fix: report unreachable domains instead of 'No meaningful changes'

Closes #10.

resolvedSnapshotAfterFallback replaces a failed lookup's snapshot with
the previous one, so alertDescriptor compared the old snapshot against
itself, found matching hashes, and the run reported 'No meaningful
changes' for a domain that was never actually reached. Nothing in the
UI or the monitoring log distinguished that from a genuine no-change.

MonitoringDomainResult now carries unreachableReason, set when the
fallback fires. It is Optional so already-persisted monitoring logs
still decode. The run summary reads 'Could not check — kept the previous
result' with the underlying error, and monitoringEvents emits a warning-
severity monitoringFailure so configured integrations hear about it
rather than seeing silence.

0 files changed, 0 insertions, 0 deletions