[Freshair129]

Summary

Recover the approved G4 diagnostic implementation left on the archived HQL2 branch (ed75e16d, strengthened by d385c608) onto current main 5eb2fa4.

• Replace preliminary optional numeric actuals with typed units and measured/unknown readings, including availability reasons and overflow.
• Report truthful clock/sampling/scope metadata; SourceScan elapsed time remains explicitly uninstrumented because materialization precedes the operator timer.
• Preserve current deterministic plan identity, result rows, budget failures, P6 authorization/snapshot boundaries, and v1 transport contracts.
• Pin the serialized vocabulary and zero/unknown/overflow distinction in regression tests; synchronize the ADR/spec/report/registry and record the recovery RCA.
• Include a separate prerequisite commit for three strict-Clippy failures reproduced on clean main: equivalent Option predicates and removal of a redundant closure drop. No lint suppression is added.

Validation

• Expanded local regression before the test-fixture follow-up: 46 test executables (library + 45 integration targets), 547 passed, 0 failed, 1 existing ignored parser child entrypoint exercised by its parent. Includes all 40 root-HQL2 targets (411 passed), 42 library tests, 30 gossip-membership tests, 10 journal-format tests, 4 journal-migration tests, 2 P4 journal-recovery tests and 48 REST tests.
• After the fixture-only follow-up: annotation ACL (12), annotation lookup (6) and G4 contract (2) pass 20/20, and strict all-target Clippy passes again in both configurations. Production source/build inputs are unchanged from the expanded run and successful builds.
• The explicit test invocation uses --locked --no-default-features --features bins --jobs 2 --lib plus the named targets above; no unfiltered stress/probe/soak suite was run locally.
• Strict all-target Clippy in both core and default/NAPI configurations: passed, with -D warnings and no lint suppression.
• Default addon/library build and standalone server build (--no-default-features --features bins --bin genesis-db-server): passed.
• Documentation validation: 244 files, 0 violations; version SSOT, cargo fmt --check and git diff --check: passed.
• Local validation is complete. Hosted checks at head c824971db49f385f48d4d8afff247a349008d965 must finish green before merge.

Compatibility and qualification

C-3 / HIGH. Core v2 diagnostic consumers must migrate from the preliminary fields to the approved typed readings. Request/envelope versions, planner identity, normal rows, persistence and v1 behavior are unchanged. No new transport route or user-database migration is introduced.

The scoped architecture review is by the implementer, not an independent review. This closes the bounded archived-code recovery only; broader G4/P8/P13, independent, device and release qualification remain open. Merge requires completed local validation and green hosted checks at the exact PR head.

Hosted incident and fixture follow-up

The first Windows job failed three annotation ACL cases in the shared database-ID setup helper with QUERY_BUDGET_EXCEEDED / query_limit. The budget dimension and underlying runner cause remain unconfirmed; a similar historical Windows budget incident exists in the P8 report.

The helper used a complete timed Values query solely to obtain fixture metadata. It now derives the expected database ID from the existing P8 public-key fingerprint contract. All annotation assertions are preserved. This follow-up changes one test helper and adds its RCA; production query budgets and engine code are unchanged. The initial failure is retained as evidence, and merge waits for fresh hosted checks at the updated head.

Final hosted qualification and merge

At head c824971db49f385f48d4d8afff247a349008d965, 49 checks succeeded and 5 conditional checks skipped; all 10 workflows completed successfully, including Tests and Mobile Build. The full Windows suite passes, including all three initially failing annotation cases. Merged normally with ancestry preserved as 7472079d37a91a81a85fc0a57824156a13481e71.