[barelyworkingcode]

Closes #439

JOURNEY_BUDGET_MS goes from 480 s to 600 s, with README and design A12 to match. The eve#424 full run on main (4bb2c32) spent 480.19 s and left add-browser-in-window BLOCKED run budget spent; the two outside-session journeys made a full run longer than the budget. No journey, assertion or timeout changes; this narrows the BLOCKED path.

Fix attempts: 1. Tokens: about 50k.

Red on main: the main harness at 4bb2c32, --epic 424: 49 PASS, add-browser-in-window BLOCKED run budget spent (journey timings summed to 480,190 ms)
Real-app check: n/a: no product code; the harness's own budget
Break-and-restore: budget cut to 500 ms, --only today-phone reported BLOCKED run budget spent for every journey; restored byte for byte (sha256 matched), the same run passed 3 of 3
Main harness vs head: main 4bb2c32 harness against the head's app (identical app code): add-browser-in-window, today-phone, cos-outside-hooks with fixtures, 5 of 5 PASS
Harness weakening: none removed or weakened; the only change is a larger shared budget, which narrows the BLOCKED path
Journey: n/a: verify runs at epic close

🤖 Generated with Claude Code

[barelyworkingcode]

Review: no blocking findings (fresh reviewer, default model; minor: FEATURES.md Notes still said 480 s, fixed in 896c729 as a logged doc amendment)
Harness weakening: none (reviewer: no assertion, check or failure path removed or weakened, no per-journey timeout changed; the larger shared budget narrows the BLOCKED path)

Two head full runs were cut short by the local Chat model slowing at 13:20 (median request about 1 s before, 8-14 s after, up to 69 s). The proof that a full run fits is the --epic 424 run on main after merge, once latency is normal. If that run still spends the budget, this PR is reverted.

[barelyworkingcode]

merge-check: merged. issue, approved, base, ci, contract, review, proof, harness, owner