follow-ups from batch review: drawer header contrast, Live heat sync, RX days race, IATA interval overflow

[dborup]

Summary

Small follow-ups found by the second review round of the issue batch (#109–#126). None blocks anything; each is a few lines plus a test. Line numbers refer to master ddae955a.

1. Nav drawer header is unreadable in the light theme (pre-existing)

.nav-drawer-header uses background: var(--surface-2, var(--surface)) (public/style.css, around line 5088), which is white in the light theme, while the title uses --nav-text, also white. "NAVIGATE" is invisible in light mode; dark mode is fine (rgb(35,35,64) background). Measured on master and on #142's branch.

• Fix: give the header the nav background (or give the title a text colour that contrasts with --surface-2), via existing CSS variables.
• Test: computed contrast between header background and title colour in both themes.

2. Live: Heat checkbox can disagree with the heat layer (from #135)

With Matrix and Heat both saved as on, turning Matrix off during page init leaves the Heat checkbox unchecked while the heat layer is shown. syncHeatToggleToMatrix(false) (public/live.js, around line 1206) only re-enables the checkbox; applyLiveControlEffects() follows heatEnabled.

• Fix: in the else branch set ht.checked = heatEnabled as well.
• Tests: this case; and a regression test for the double-listener bug fix(live): wire every persisted view toggle before Live init awaits #135 fixed, which only shows when Live is left before init's awaits finish (master: 2 listeners on 8 of 9 toggles, one click writes localStorage twice). The current "three SPA round trips" test also passes on master.

3. RX Coverage: a stale days response can win (pre-existing)

Switching days quickly (setDays, public/rx-coverage.js, around line 211) starts new coverage/leaderboard fetches without invalidating the previous ones, so an older, slower response can render last.

• Fix: a request generation (as fix(rx-coverage): honour configured defaults and save an independent viewport #136 does for mount/viewport), checked before rendering coverage and the leaderboard.
• Test: delay the first response, switch days, release it, assert the newer data stays.

4. Ingestor: huge iataWarnIntervalSec overflows (from #134)

IATAWarnInterval() (cmd/ingestor/iata_drop_warn.go, around line 55) returns time.Duration(c.IATAWarnIntervalSec) * time.Second; a very large value (e.g. 1e10) overflows to a negative duration, so every dropped message is logged, which is the log flood the throttle exists to prevent.

• Fix: clamp to a sane maximum (e.g. 24 h) and document it.
• Tests: overflow value, and the untested oldest optimisation with staggered timestamps (two mutants survived the current tests).

Acceptance criteria

• Each item has a test that fails on master and passes with the fix.
• No behaviour change beyond the four items; existing suites stay green.

[dborup]

5. Analytics warm-up window vs. frontend retry (from fix(analytics): recompute cached snapshots after the complete startup load #145 / fix(analytics): recompute cached snapshots after the complete startup load #116 review). Since fix(analytics): recompute cached snapshots after the complete startup load #145 the RF/Topology/Channels 503 warm-up lasts until the whole background load has finished (still capped by the 60 s force-open), while the frontend only retries for about 30 s. On a slow startup a user can therefore see the error state before the data becomes available. Either let the frontend keep retrying (honouring Retry-After) until the server stops answering 503, or show a "still loading" state instead of an error. Test with a delayed background load.

[dborup]

Items 1–4 are fixed by #169 (merged as 2c7fbd6c):

1. The drawer header now uses --nav-bg2. Title contrast in the default light theme went from 1.00:1 to 17.06:1.
2. Live: the Heat checkbox now follows Heat when Matrix goes off. A regression test for the fix(live): wire every persisted view toggle before Live init awaits #135 double-listener bug was added.
3. RX Coverage: coverage and leaderboard render only the latest request. This covers days, pan and rx switches.
4. Ingestor: iataWarnIntervalSec is capped at 86400 s (24 h). This is documented in config.go and config.example.json.

Still open: item 5 (analytics warm-up vs. frontend retry). It moves to #172, together with three small P3 findings from the #169 review. This issue stays open until #172 is done.

[dborup]

Note: this issue was closed automatically by the merge of #169, because the PR title starts with "fix: #150", which GitHub reads as a closing keyword. Items 1–4 are fixed by #169. Item 5 (analytics warm-up vs. frontend retry) is still open and is tracked in #172 (being fixed in #175), so keeping this one closed is fine.