…learing

Fixes the critical issue where users are randomly logged out due to TokenOrchestrator clearing tokens for ANY non-network error, including transient AWS service issues.

Resolves #14534

The `handleErrors()` method was clearing tokens for all errors except NetworkError, causing users to be logged out during:
- AWS service errors (500s)
- Rate limiting (TooManyRequestsException)
- Temporary service unavailability
- Other transient errors that should be retried

Modified token clearing logic to only clear tokens for definitive authentication failures that require re-authentication. Transient errors now preserve tokens, allowing for retry without forcing users to log in again.

- Removed overly broad token clearing condition
- Added `isAuthenticationError()` helper to identify auth failures
- Only clears tokens for errors that definitively indicate invalid/expired tokens
- Preserves tokens for all transient/retryable errors

- `NotAuthorizedException` - Refresh token expired/invalid
- `TokenRevokedException` - Token explicitly revoked
- `UserNotFoundException` - User no longer exists
- `PasswordResetRequiredException` - Password reset required
- `UserNotConfirmedException` - Account not confirmed

- `InternalErrorException` - AWS service errors
- `TooManyRequestsException` - Rate limiting
- `ThrottlingException` - Request throttling
- `ServiceUnavailable` - Temporary service issues
- `NetworkError` - Network connectivity issues
- Any other transient or unknown errors

- Added comprehensive test coverage for all error scenarios
- Tests verify correct token clearing behavior for each error type
- All existing tests pass

This fix ensures users remain logged in during temporary AWS service issues while still properly handling genuine authentication failures.