What changed

Fixed a crash that could happen when the Google Play billing service disconnects unexpectedly (e.g., during a Play Store update or when the OS kills the billing process).

Technical Context

• Root cause: After BillingClientConnectionProvider's 5 retry attempts are exhausted, the BillingException propagates to BillingDataRepo.connectionProvider which had a .catch { throw it } before .replayingShare(scope). The re-throw causes shareIn's internal coroutine to fail with an unhandled exception in the SupervisorJob-based AppScope, crashing the app.
• Fix: Replaced the .catch { throw it } with .retryWhen that retries indefinitely with 60s backoff. This keeps the SharedFlow alive — when the billing service returns, the connection re-establishes automatically.
• The existing upgradeInfo flow already has onStart fallback values (grace period), so the UI isn't blocked during retry waits.