fix: prevent IDB write-stall from blocking trip page and sync loop

clearAll() now clears all tables in a transaction instead of calling
offlineDb.delete(), which triggered our versionchange handler and put
Dexie into a broken write state for the rest of the session.

tripRepo.get() gets the same 2 s timeout guard as list() so a stalled
IDB read no longer freezes the trip splash screen.

_doSync wraps each syncTrip() in a 30 s per-trip timeout so a single
stalled write transaction cannot prevent the loop from advancing to
subsequent trips.

client/src/repo/tripRepo.ts

@@ -48,7 +48,10 @@ export const tripRepo = {
   },
 
   async get(tripId: number | string): Promise<{ trip: Trip; refresh: TripRefresh }> {
-    const cached = await offlineDb.trips.get(Number(tripId))
+    const cached = await Promise.race([
+      offlineDb.trips.get(Number(tripId)).catch(() => undefined),
+      new Promise<undefined>(resolve => setTimeout(() => resolve(undefined), 2000)),
+    ])
 
     const refresh: TripRefresh = (async () => {
       try {