docs(project): drop EA framing from umbrella plan/spec

EA shipped without this work, so the umbrella no longer has a deadline.
Replace the pre-EA/post-EA sequencing with deadline-free framing: S1+S2
are the must-ship core, S3 is additive and can land independently. The
default-namespace fallback is reframed as a backward-compatibility
mechanism for existing consumers rather than EA protection. Removes the
EA-timeline constraint/risk and renumbers the surrounding items.

Signed-off-by: Will Madden <madden@prisma.io>

Author: wmadden

projects/target-extensible-ir-namespaces/plan.md

@@ -2,11 +2,11 @@
 
 **Spec:** [`projects/target-extensible-ir-namespaces/spec.md`](./spec.md)
 **Linear Project:** [Target-Extensible IR + Namespaces](https://linear.app/prisma-company/project/target-extensible-ir-namespaces-fd69eff8aec6)
-**Purpose** _(from spec)_: Make first-class namespaces and target-extensible IR usable for the downstream Supabase integration. The IR gains a pack-contributed entity-kind mechanism (proven by Postgres enum migrating off the framework-shared `types` slot); runtime SQL and the DSL/ORM surfaces qualify identifiers and dispatch through default-namespace fallback so EA users on single-namespace contracts experience zero breakage; the explicit namespace-aware surface (`db.sql.auth.user`) ships post-EA as purely additive work.
+**Purpose** _(from spec)_: Make first-class namespaces and target-extensible IR usable for the downstream Supabase integration. The IR gains a pack-contributed entity-kind mechanism (proven by Postgres enum migrating off the framework-shared `types` slot); runtime SQL and the DSL/ORM surfaces qualify identifiers and dispatch through default-namespace fallback so existing single-namespace consumers experience zero breakage; the explicit namespace-aware surface (`db.sql.auth.user`) lands later as purely additive work.
 
 ## At a glance
 
-Three units, single stack thread: **S1 (sub-project) → S2 (slice) → S3 (slice)**. S1 + S2 are pre-EA must-ship; S3 is post-EA additive. No parallelisation at the umbrella level — each unit gates the next on substrate, not just on convention. The umbrella plan delegates execution detail to S1 (now closed; durable decisions in [ADR 221](../../docs/architecture%20docs/adrs/ADR%20221%20-%20Contract%20IR%20two%20planes%20with%20uniform%20entity%20coordinate%20and%20pack-contributed%20entity%20kinds.md)) and to S2/S3 slice specs which will be authored via `drive-specify-slice` at pickup time.
+Three units, single stack thread: **S1 (sub-project) → S2 (slice) → S3 (slice)**. S1 + S2 are the must-ship core; S3 is additive and can land independently. No parallelisation at the umbrella level — each unit gates the next on substrate, not just on convention. The umbrella plan delegates execution detail to S1 (now closed; durable decisions in [ADR 221](../../docs/architecture%20docs/adrs/ADR%20221%20-%20Contract%20IR%20two%20planes%20with%20uniform%20entity%20coordinate%20and%20pack-contributed%20entity%20kinds.md)) and to S2/S3 slice specs which will be authored via `drive-specify-slice` at pickup time.
 
 ## Composition
 
@@ -26,13 +26,13 @@ Three units, single stack thread: **S1 (sub-project) → S2 (slice) → S3 (slic
 
 **Validation gate.** Each S1 slice has its own validation gate (see S1's plan). S1 closes when all six slices land and S1's PDoD1-PDoD11 are met.
 
-**Pre/Post-EA.** Pre-EA must-ship.
+**Priority.** Must-ship core (closed).
 
 #### S2 — runtime SQL qualification + default-namespace DSL/ORM fallback
 
 **Unit type:** Slice (single PR).
 
-**Purpose.** Close the runtime loop on PR #534's namespace IR: runtime SQL emits namespace-qualified identifiers; DSL/ORM reads through per-family default-namespace fallback (`'public'` for Postgres; `'__unbound__'` for Mongo/SQLite) so legacy query code keeps working unchanged. The fallback is the load-bearing EA-protection mechanism.
+**Purpose.** Close the runtime loop on PR #534's namespace IR: runtime SQL emits namespace-qualified identifiers; DSL/ORM reads through per-family default-namespace fallback (`'public'` for Postgres; `'__unbound__'` for Mongo/SQLite) so legacy query code keeps working unchanged. The fallback is the load-bearing backward-compatibility mechanism.
 
 **Scope.** ~10-12 files: family façade hardcoded `defaultNamespace` constant; DSL `Db<C>` type-resolution path; ORM accessor type-resolution path; runtime SQL identifier-qualification at the relational-core / runtime layers. Slice spec authored via `drive-specify-slice` at pickup time; lands at `projects/target-extensible-ir-namespaces/slices/runtime-qualification/spec.md`.
 
@@ -49,7 +49,7 @@ Three units, single stack thread: **S1 (sub-project) → S2 (slice) → S3 (slic
 - `pnpm lint:deps`
 - **Project-specific check:** the demo app's `examples/prisma-next-demo/src/queries/*.ts` files compile and run unchanged after S2 lands. (Surfaces FR5 satisfaction.)
 
-**Pre/Post-EA.** Pre-EA must-ship.
+**Priority.** Must-ship core.
 
 #### S3 — explicit namespace-aware DSL/ORM surface
 
@@ -72,7 +72,7 @@ Three units, single stack thread: **S1 (sub-project) → S2 (slice) → S3 (slic
 - `pnpm lint:deps`
 - **Project-specific check:** demo app migrates one query from `db.sql.user` to `db.sql.public.user` (explicit-namespace form) and both forms compile + run.
 
-**Pre/Post-EA.** Post-EA additive.
+**Priority.** Additive (can land independently).
 
 ### Parallel groups
 
@@ -97,36 +97,35 @@ Within S1, S1's own plan defines parallelisation opportunities (Slices 4 + 5 are
 
 | Project-DoD | Delivered by |
 |---|---|
-| **PDoD1.** All units delivered or deferred | S1 + S2 (pre-EA) + S3 (post-EA, or deferred to a sibling initiative) |
+| **PDoD1.** All units delivered or deferred | S1 + S2 + S3 (additive; or deferred to a sibling initiative) |
 | **PDoD2.** Multi-namespace contract authorable + emittable + queryable end-to-end | S1 (authoring + emission) + S2 (queryable) |
 | **PDoD3.** Pack-contributed entity-kind substrate exercised end-to-end (Postgres enum migration) | S1 (inherits its own PDoD3) |
 | **PDoD4.** Runtime SQL identifier-qualification | S2 |
-| **PDoD5.** EA users on default-namespace contracts experience zero query-API breakage | S2 (default-namespace fallback) + verified across S1+S2 landings |
+| **PDoD5.** Existing consumers on default-namespace contracts experience zero query-API breakage | S2 (default-namespace fallback) + verified across S1+S2 landings |
 | **PDoD6.** Long-lived ADRs migrated to `docs/architecture docs/adrs/` | Close-out task; lifts ADRs from S1 + any S2/S3 ADRs |
 | **PDoD7.** Linear Project marked Completed | Close-out task (auto via PR-merge integration) |
 | **PDoD8.** Rolled-up project folders archived; umbrella folder deleted | Close-out task |
 | **PDoD9.** Repo-wide references stripped | Close-out task |
 
 ## Risks + open questions
 
-1. **EA timeline slip.** A2 in the spec is the binding constraint. Honest budget for pre-EA scope is ~12-15 days (S1: 10-12 + S2: 2-3); "few days" framing was unrealistic. Working position is to slip pre-EA by 1-2 weeks rather than ship breaking IR changes into EA. If the slip is unacceptable to the team, the conversation re-opens at umbrella level — options are (a) ship without S1's IR reshape and accept breaking changes for EA users when S1 lands later, (b) compress S1 via aggressive parallelism (compression-sprint plan, previously discarded), or (c) defer S2's namespace-qualification too and ship EA on the unchanged surface with namespaces deferred to the next minor.
-2. **S1 internal slice falsifications cascade to umbrella.** *(Resolved — S1 closed; durable decisions in [ADR 221](../../docs/architecture%20docs/adrs/ADR%20221%20-%20Contract%20IR%20two%20planes%20with%20uniform%20entity%20coordinate%20and%20pack-contributed%20entity%20kinds.md).)* S1's assumptions held without forcing an umbrella-level re-sequencing of S2 / S3.
-3. **A4 default-namespace fallback sufficiency.** If EA users start writing multi-namespace contracts before S3 lands, the flat DSL surface becomes inadequate and S3's post-EA window compresses. Mitigation: EA release notes communicate the namespace-aware surface as "coming soon."
-4. **Linear audit surfaces unexpected scope.** The 21 Linear tickets currently in the project may include work that wasn't accounted for in this umbrella's three-unit composition. If audit surfaces real-scope work the umbrella missed, the plan re-opens.
+1. **S1 internal slice falsifications cascade to umbrella.** *(Resolved — S1 closed; durable decisions in [ADR 221](../../docs/architecture%20docs/adrs/ADR%20221%20-%20Contract%20IR%20two%20planes%20with%20uniform%20entity%20coordinate%20and%20pack-contributed%20entity%20kinds.md).)* S1's assumptions held without forcing an umbrella-level re-sequencing of S2 / S3.
+2. **A3 default-namespace fallback sufficiency.** If consumers start writing multi-namespace contracts before S3 lands, the flat DSL surface becomes inadequate and S3 rises in priority. Mitigation: document the namespace-aware surface as a planned addition so consumers know explicit navigation is coming.
+3. **Linear audit surfaces unexpected scope.** The Linear tickets in the project may include work that wasn't accounted for in this umbrella's three-unit composition. If audit surfaces real-scope work the umbrella missed, the plan re-opens.
 
 ## Sequencing visualisation
 
 ```text
 PR #534 ✓ (merged)
    │
    ▼
-S1 — contract-ir-planes (sub-project, 6 slices internally, ~10-12 days)
+S1 — contract-ir-planes (sub-project, 5 merged slices)   ✓ CLOSED
    │
    ▼
-S2 — runtime-qualification (slice, ~2-3 days)
+S2 — runtime-qualification (slice)                       → next
    │
-   ▼   ←  EA cut here
-S3 — explicit-namespace-dsl (slice, ~2-3 days, additive, post-EA)
+   ▼
+S3 — explicit-namespace-dsl (slice, additive)
    │
    ▼
 Downstream: Supabase integration consumes this substrate