fix: prevent staking dashboard infinite loading

- Add maxAttempts parameter to retryUntilDone (default unlimited for
  backward compat), use 5 retries for staking stats fetch
- Catch fetchStakingStats failure in dashboard update system and
  fallback to empty stats instead of hanging forever
- Restore stale scope detection in ComputationalCache so cancelled
  aggregate scopes are recreated instead of returning stale entries

feature-staking-impl/src/main/java/io/novafoundation/nova/feature_staking_impl/data/dashboard/network/stats/StakingStatsDataSource.kt

@@ -39,7 +39,7 @@ class RealStakingStatsDataSource(
         stakingAccounts: StakingAccounts,
         stakingChains: List<Chain>
     ): MultiChainStakingStats = withContext(Dispatchers.IO) {
-        retryUntilDone {
+        retryUntilDone(maxAttempts = 5) {
             val globalConfig = globalConfigDataSource.getGlobalConfig()
             val chainsByEndpoint = splitChainsByEndpoint(stakingChains, globalConfig)
 

feature-staking-impl/src/main/java/io/novafoundation/nova/feature_staking_impl/data/dashboard/network/updaters/StakingDashboardUpdateSystem.kt

@@ -127,9 +127,16 @@ class RealStakingDashboardUpdateSystem(
             .onEach { latestOffChainSyncIndex.value = it.index }
             .throttleLast(offChainSyncDebounceRate)
             .mapLatest { (index, stakingAccounts) ->
+                val stats = runCatching {
+                    stakingStatsDataSource.fetchStakingStats(stakingAccounts, stakingChains)
+                }.getOrElse {
+                    Log.d("StakingDashboardUpdateSystem", "Failed to fetch staking stats after retries", it)
+                    emptyMap()
+                }
+
                 MultiChainOffChainSyncResult(
                     index = index,
-                    multiChainStakingStats = stakingStatsDataSource.fetchStakingStats(stakingAccounts, stakingChains),
+                    multiChainStakingStats = stats,
                 )
             }
     }