chore: CRDT protocol micro-optimizations

[pravusjif]

What

Low-risk data-structure optimizations on the CRDT hot path (scene ↔ renderer messaging), one PR of a series based on a CRDT pipeline audit:

• Merge matrix → flat array — WorldSyncCommandBuffer's MERGE_MATRIX dictionary (probed per touched (entity, component) per batch, with a HashCode.Combine hash) replaced with a flat CRDTReconciliationEffect[25] indexed by (first * 5) + last.
• Presized protocol state dictionaries — CRDTProtocol's outer/inner PooledDictionary buckets get explicit capacities, avoiding rehash cascades (each rehash re-rents + copies the pooled backing arrays) during scene-load PUT storms.
• Single-probe entity deletion — DeleteEntity used TryGetValue + Remove (two probes) per component bucket per deleted entity; now Remove(key, out value).
• Packed outgoing message key — (entity, componentId) packed into a long so the pending-messages dictionary uses the devirtualized default comparer instead of a comparer class with HashCode.Combine per probe.
• LWW outgoing messages committed on creation — previously each outgoing PUT/DELETE probed the state dictionaries twice: once in CreateLwwMessage (timestamp) and again in EnforceLWWState after serialization (full LWW reconciliation whose outcome is predetermined for local messages). CreateAndCommitPutMessage/CreateAndCommitDeleteMessage do it in one pass; EnforceLWWState is removed.
• Micro: CRDTEntity now overrides GetHashCode/Equals(object) (removes the default-comparer boxing footgun); the per-message no-op ProcessPendingMessage delegate is skipped in the base EngineAPIImplementation (only the SDK-observables subclass provides one); the per-message AUTHORITATIVE_PUT_COMPONENT deserialization log (editor/verbose-only, interpolated string per message) is removed — shout if it's still wanted for debugging.

Why

These are the per-message costs of every scene tick, multiplied by scene count. None of the changes alters protocol semantics: LWW commit-on-creation produces byte-identical state to the previous create + enforce pair (local messages always win by construction).

QA / verification

• Affected assemblies + DCL.EditMode.Tests compile clean locally (dotnet build, 0 errors).
• Existing EditMode suites cover the touched code (CRDTProtocolShould, CRDTShould protocol-conformance files, WorldSyncCommandBufferShould, OutgoingCRTDMessagesProviderShould — the last updated for the renamed/packed APIs). Please rely on the CI EditMode run for execution; a local Test Runner run was not possible (editor lock).
• No QA steps beyond a smoke test of any SDK7 scene: components add/modify/delete, entity deletion, and SDK observables (onEnterScene/onSceneReady) still work.

🤖 Generated with Claude Code

[github-actions]

New build in progress, come back later!

[m3taphysics]

PR #8967, run #27390644825

Builds: Windows change, Windows baseline, macOS change, macOS baseline

How to read this table

• Each build is measured 3 times. The values are the median, and (min–max) is the lowest and highest of those runs — a wide range means the metric is noisy and small differences are not trustworthy.
• Δ is Change minus Baseline (a negative Δ means Change is faster).
• 🟢 faster / 🔴 slower — a real difference: larger than both 3% and the run-to-run range.
• ⚪ within noise — the difference is smaller than how much the build varies between its own runs, so it cannot be told apart from random variation. Treat it as no change.
• Exceptions logged — exceptions found in the run logs; more than the baseline is flagged 🔴 even when frame times look fine. The Exception breakdown under each table groups them by the explorer's report category and exception type.
• A run that logged unusually many exceptions (at least 10 and 5× the median of its build's runs — e.g. a service was down during it) is excluded from all numbers and called out under the table.

Framework 13 i7

Metric	Baseline	Change	Δ	Result
Samples	2701 (×3)	2701 (×3)
CPU average	33.3 ms	33.3 ms	-0.0 ms	⚪ within noise
CPU 1% worst	33.8 ms (33.7–34.0)	33.5 ms	-0.3 ms	⚪ within noise
CPU 0.1% worst	37.1 ms (36.0–37.2)	34.0 ms (33.9–34.2)	-3.2 ms	🟢 9% faster
GPU average	7.5 ms (7.5–7.6)	8.0 ms (7.5–8.0)	0.5 ms	⚪ within noise
GPU 1% worst	19.7 ms (19.2–20.0)	19.5 ms (19.4–20.2)	-0.2 ms	⚪ within noise
GPU 0.1% worst	24.8 ms (24.1–26.8)	26.7 ms (25.4–28.6)	1.9 ms	⚪ within noise
Exceptions logged	6	6	0	⚪ none new

Exception breakdown

Exception	Baseline	Change
[ENGINE] NullReferenceException	6	6

[github-actions]

Warnings count reduced: 48555 => 48552