test(web): compare work log update cost against a from-scratch derive

[rynfar]

What Changed

apps/web/src/session-logic.test.ts: the 20,000-activity work-log benchmark no longer asserts a 100 ms wall-clock bound. It now measures, in the same process, an update over cached rows and a from-scratch derivation of the same rows, takes the best of three rounds for each, and asserts the update is under half the from-scratch time.

Why

Since #99 moved CI to GitHub-hosted standard runners (4 vCPU, four Vitest workers sharing them), the old bound failed on two of the first three runs — 100.5 ms and 107.5 ms — for work that takes ~4 ms on a quiet laptop. It was measuring runner contention, not deriveWorkLogEntries, and every unrelated PR (#100, #101) had to be rerun to get green.

The optimization the test was written for (upstream #8006, cherry-picked as b49ad7918) is a per-activity WeakMap cache so an update does not re-derive every row. Measured here:

	this laptop (M5 Pro)
from-scratch derive, 20k rows	30–46 ms
update over cached rows, 20k+1	4–8 ms
derive with cache misses (the pre-#8006 path)	29–34 ms

So the ratio the test now asserts is ~0.13–0.27 with the cache and ~1.0 without it; a bound of 0.5 has margin on both sides and is independent of how fast or contended the machine is. The existing identity test two cases above (appendedEntries[1] is initialEntries[1]) still pins exact row reuse deterministically.

Verified: vp test run apps/web/src/session-logic.test.ts — 82 passed; vp run -F @t3tools/web typecheck clean.

Checklist

• This PR is small and focused
• I explained what changed and why
• I included before/after screenshots for any UI changes (no UI change)
• I included a video for animation/interaction changes (no motion change)

Model: Claude Fable 5 · Harness: Claude Code

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}

[github-actions]

Thread transfer impact

✅ Thread transfer remains within every enforced ceiling.

Provider	Metric	Main baseline	This PR	Impact	PR ceiling
Codex	Total thread wire	13.3 KiB	13.3 KiB	−1 B (−0.0%)	15.1 KiB	✅
Codex	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Codex	Live turn WebSocket wire	6.4 KiB	6.4 KiB	−6 B (−0.1%)	7.8 KiB	✅
Codex	Live turn WebSocket decoded	55.5 KiB	55.5 KiB	0 B (0.0%)	66.4 KiB	✅
Codex	Live turn messages	9	9	0 (0.0%)	21	✅
Claude	Total thread wire	13.3 KiB	13.3 KiB	−1 B (−0.0%)	15.1 KiB	✅
Claude	Thread snapshot wire	6.9 KiB	6.9 KiB	+5 B (+0.1%)	7.3 KiB	✅
Claude	Live turn WebSocket wire	6.4 KiB	6.4 KiB	−6 B (−0.1%)	7.8 KiB	✅
Claude	Live turn WebSocket decoded	56.3 KiB	56.3 KiB	0 B (0.0%)	66.4 KiB	✅
Claude	Live turn messages	9	9	0 (0.0%)	21	✅

Baseline: 486065e · PR result: a2b1d18 · Source CI: success

Scenario and decoded snapshot size

10 historical turns, 5 command tools per turn, 878.9 KiB retained MCP result per historical turn, and a 1.05 MiB retained result in the measured turn.

• Codex decoded thread snapshot: 109.4 KiB
• Claude decoded thread snapshot: 110.0 KiB

Updated in place by a trusted workflow. PR artifacts are strictly validated and never executed.