[NaNIsaac]

Tier: low (test timeout only)

• wait_until waits up to 30 s instead of 10 s for the explorer listing. No source change; on an idle machine the tests take as long as before, because the wait ends as soon as the condition holds.
• Why: GUI filter tests are timing-dependent and fail under load #8 failed again after test(gui): delete windows on teardown and wait for the explorer listing #22 (1 of 20 runs on main under 12 busy-loop processes, 2026-10-23), with the listing empty but root path, name filters and the filter timer all as the test set them. A controlled rerun registered in advance separated the two readings: 20/20 with one core left free (11 workers, 10 s wait; runs took about 5 s) and 20/20 with all 12 cores busy and a 30 s wait (runs took 46 s median, 58 s at most). So the failures were QFileSystemModel's worker thread starved past 10 s, not leftover state.
• The registered verdict for both 20/20 was "gathering delay under CPU saturation: raise the test timeout and close GUI filter tests are timing-dependent and fail under load #8".

run (2026-10-23, QoraDFXM main, 6 filter tests, 20 runs each)	load	wait	passed	run time median / max
B16b, main 1e27ef4	12 busy processes	10 s	19/20 (1 empty listing)	30-40 s
B16b, #26 a5dd6eb	12 busy processes	10 s	19/20 (1 empty listing)	30-40 s
B16c (a), main 1e27ef4	11 busy processes	10 s	20/20	4.7 / 13.9 s
B16c (b), main 1e27ef4 + this change	12 busy processes	30 s	20/20	46 / 58 s

B16b ran with the Everything indexer and Defender active in some snapshots; B16c had nothing else above 0.6 core.

Fixes #8.

Local: the 6 filter tests (-k "filter or escape or clearing") 20/20 under 12 busy processes with this change. The full GUI suite was not rerun locally (another long job holds the machine); CI runs it.