Record two traps: the hygiene gates targeted runs miss, and what a present-wait robot timeout means

[samoylenkodmitry]

Two TIME_WASTERS entries, both paid for in full this afternoon.

The hygiene gates a per-crate run never reaches. apps/desktop-demo/tests/source_hygiene_aliases.rs scans every crate's sources, so a /tmp/ literal in a pure in-memory fixture takes main red while a targeted -p <crate> gate set stays green — and a sandboxed reviewer's just test can die at the socket-bound cranpose-services::peer tests before reaching it. Both happened at once on #495, and four sessions then fixed the same line in parallel (#503, #504, #506, #507).

A waiting_for_present=true robot timeout is a host-load symptom. wait_for_idle budgets loop iterations rather than wall clock, so a loaded machine fails healthy tests; its own diagnostics (needs_update=false, has_animations=false) say composition already settled. The same test passed in isolation and the full 121-test suite passed on a re-run of the same commit. The underlying harness fix is tracked separately.

🤖 Generated with Claude Code