GUI: transient tests, per-fixture results, passed panels that go

[jp-pino]

For the battery station's new start flow (BluEye-Robotics/P2-functional-tests#244). Its first column is a test that asks which slot to start next. When all four slots are running, that column should go, so the station shows at most four columns.

Changes

• Transient tests. A test started with htf.Test(..., transient=True) (it ends up in the record's metadata) has its panel removed as soon as it completes. TestState.transient comes from test_record.metadata.transient.

• Each fixture keeps its result. A new test now replaces only the completed tests of its own name. Before, any new test cleared every completed panel. The battery station restarts its first column whenever a slot frees up, and under the old rule a slot's PASS/FAIL would have vanished at that moment. Now a slot's result stays until that slot starts again. A station with a single test name (every other station) behaves exactly as before.

• Passed panels can go after a delay. StationServer(..., hide_passed_after_s=120) passes the delay to the GUI in its config (ConfigService.hidePassedAfterMs). With it set, a passed test's panel is removed that long after it completes. A test that failed, errored or was aborted keeps its panel until a test of its name starts again. When it isn't set, nothing changes.

• Transient tests stay out of History. A transient test's final record isn't added there; otherwise every one showed up as an UNKNOWN_DUT pass.

Testing

• Two new cases in station.service.spec.ts:
  • a transient test is dropped when it completes while a slot keeps running;
  • a completed "Slot 1" survives "Slot 2" starting and is replaced when "Slot 1" starts again.
• More cases in station.service.spec.ts:
  • a pass goes at 120 s while a FAIL next to it stays;
  • without the setting, a pass stays;
  • a transient test's final record doesn't reach History.
• ng test in headless Chrome: 43 of 43 pass.
• test/output/servers/station_server_test.py: the server renders hide_passed_after_s into its page, as null when it's not set. 6 of 6 pass.
• Checked live with the battery station on the bms_tools simulator (BluEye-Robotics/P2-functional-tests#244).
• dist/ rebuilt with Node 24; only the main bundle changes.

🤖 Generated with Claude Code

[coveralls]

Coverage Report for CI Build 38723482313

Coverage increased (+1.3%) to 67.593%

Details

• Coverage increased (+1.3%) from the base build.
• Patch coverage: 1 of 1 lines across 1 file are fully covered (100%).
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	7739
Covered Lines:	5231
Line Coverage:	67.59%
Coverage Strength:	3.38 hits per line

---

💛 - Coveralls