ragleap-observability v0.8.0 release notes

Opt-in availability SLO alerts for ragleap-observability.

slo.enabled (off by default): fast-burn (1h/5m, 14.4x) and slow-burn (6h/30m, 6x) alerts on the Prometheus up metric for slo.availability.jobs (default ragleap-postgres). The 99.5% errorBudget is a placeholder, not a measured figure.
Fix: rule_files now loads even with alertmanager.enabled=false. Before, PostgresExporterDown was silently inactive in that case.
promtool unit tests (tests/promtool/run.sh) and a CI job, slo-rules-test.

Verified: default helm template unchanged; promtool check and test pass; breaking each threshold made the tests fail.
Not verified: evaluation on a live Prometheus, and alert delivery through AlertManager. Rule logic only, on simulated data (the earlier kind cluster no longer exists). Measures target reachability, not user experience: the app exports no request metrics yet. A target that vanishes entirely does not trigger the alerts (no absent() rule yet).

pip install ragleap-observability==0.8.0