Expose email outbox observability metrics

[capybaracplusplus]

Summary

• expose bounded email outbox delivery and persistence counters through the existing private userver monitor endpoint
• record complete claimed-job processing duration as a histogram
• keep observability behind an application port and document metric names, labels, and alert intent
• add a functional contract test for every fixed label value and the duration histogram

Why

HTTP and PostgreSQL metrics already describe transport and database behavior, but they cannot show whether email jobs are sent, retried, moved to dead state, or discarded after lease loss. These signals are required for the Prometheus/Grafana platform dashboard without adding user identifiers or other high-cardinality labels.

Validation

• cmake --preset debug
• cmake --build build-debug --parallel 1
• ctest --test-dir build-debug --output-on-failure
• cmake --preset release
• cmake --build build-release --parallel 1
• git diff --check