Resi Cues 1.0.2: a failed Resi login holds off instead of retrying every second

A failed login left no session, so the next 1s poll signed in again, and the
next, for ever. Resi answers that with 429s, and since the retries never let
up the account never got out of them: at North Attleboro the decoder picker
stayed empty with "Resi login failed (429)". The picker made it worse by
building a fresh client -- and signing in -- on every admin page load.

- After a failed login, further attempts wait 15s, doubling to 5min (or as
  long as Resi's Retry-After asks); calls in between fail at once with the
  reason and how long until the next try.
- The status says what happened: rate-limited, password rejected, or Resi
  unreachable.
- The decoder picker reuses the running module's client when the login
  matches, otherwise one client per login, so it shares the hold-off.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

modules/resi-cues/module.json

@@ -1,7 +1,7 @@
 {
   "id": "resi-cues",
   "name": "Resi Cues",
-  "version": "1.0.1",
+  "version": "1.0.2",
   "proddash": ">=1.4.0",
   "description": "Time until the next Resi cue point, on the board and as a timer",
   "provides": ["timers"],