Merge train 4843-19b (CI only; not for merge)

[erikdarlingdata]

Merge train, not for merge. CI runs once on dev plus the squash of each PR below, in this order. When it is green, each is squash-merged by itself in the same order, dev's tree is checked against this branch's, and this PR is closed unmerged.

Part of #4966 and #5158.

PRs in this train, in order

• MCP: Darling's blocking stats and PostgreSQL deadlock and log event tools say where their data starts (part of Say where the data starts on every time-ranged tab, in Lite and both Darling viewers #4966) (MCP: Darling's blocking stats and PostgreSQL deadlock and log event tools say where their data starts (part of #4966) #5205)
• The store accepts a known plan digest for a row and keeps that plan alive (inert, query_stats renders and ships the same ~200 plans every minute: 47x more plan XML than is distinct, and the reason its slow runs are slow #5158) (The store accepts a known plan digest for a row and keeps that plan alive (inert, #5158) #5174)

Train head fcd6654b7, tree 5736e2e65, base dev 7aca8830a. Neither member moves the tools/list budget.

Two members were dropped from earlier builds of this train, each for a reason worth recording. One builds on the second member here and still carried that member's older commits, so its squash would have reverted a fix that the second member lands — its author is re-merging it and verifying the file byte-for-byte against dev. The other failed one of its own aliasing-hardening tests in CI with the refusal message null, meaning nothing was refused; that test is absent on dev, so it is new and attributable, and it had skipped on its author's Windows VM because that session cannot create a symbolic link, so a 23,640-passing run had never exercised it.

Local merged-tree checks: 1,793 tests, 0 build warnings; the store-upgrade class at 9 and the hygiene class, both matching dev at this commit. The repository-source scan over 402 classes reports no new failure, and the live-store hygiene guard — which cannot execute on macOS — was emulated against this merged tree and reports 0 offenders.

[erikdarlingdata]

Train landed. Its CI run was green with zero failing tests in every job - the second train tonight with no failures anywhere.

• MCP: Darling's blocking stats and PostgreSQL deadlock and log event tools say where their data starts (part of #4966) #5205 squashed to 912ac18a2
• The store accepts a known plan digest for a row and keeps that plan alive (inert, #5158) #5174 squashed to a923f679b

dev is now a923f679b and its tree matches this branch's exactly, which is the check that proves the squashes reproduced the tested tree. Closing unmerged, as a train PR always is.

Two shards ran at or past the 15-minute watch line (PG(1) 15.8m, PG(2) 16.0m). Nothing failed, but a slower runner turns that into a timeout red rather than a test failure, so it is worth watching on the next few trains.