fix: align mixin stab list with desugared IR order in IncludeD

[Kamirus]

build_stabs for IncludeD produced [letP_flex; imports...; mixin_decs...] while dec' emits [imports...; letP; mixin_decs...], so List.map2 stabilize stabs ds paired the letP stab with the first import dec and shifted each import stab onto the next dec.

Benign today (every stab in that segment is Some Flexible and stabilize is identity on Flexible), but a latent footgun the moment any of those stabs becomes anything other than Flexible.

Made with Cursor

[github-actions]

Cursor AI review

👍 APPROVE — looks safe to merge

Category	Assessment	Details
Summary	✅	Reorders build_stabs for IncludeD so stab list matches dec' IR order [imports; letP; mixin decs] instead of [letP; imports; mixin decs].
Code Quality	✅	Minimal 4-line reorder plus an invariant comment; no new helpers.
Consistency	✅	Matches existing build_stabs / dec' split and OCaml list style in desugar.ml.
Correctness	✅	List.map2 stabilize stabs ds in build_actor now pairs each stab with the matching IR dec; mixin-dec stabs were already aligned at the tail.
Tests	✅	No .mo/.ok change needed: affected prefix stabs are hardcoded Flexible and stabilize is a no-op on them today; existing mixin run-drun tests still cover the path.
Changelog	✅	No user-visible behavior change; internal ordering fix only.

Verdict

Decision: APPROVE
Risk: Very Low
Reason: The diff only fixes a latent stab/dec misalignment in the IncludeD prefix where every stab is currently Flexible, so runtime behavior is unchanged while the invariant now matches dec'. No compiler-surface, codegen, or RTS impact warrants escalation.

---

Generated for commit a8c0420

[github-actions]

Comparing from e7c78d7 to a8c0420:
The produced WebAssembly code seems to be completely unchanged.
In terms of gas, no changes are observed in 5 tests.
In terms of size, no changes are observed in 5 tests.