docs: document the metric_stability test

[joostboon]

Documents elementary.metric_stability, the new test added in dbt-data-reliability#1050. Opened here rather than in the package repo per review feedback on that PR: a lone markdown file in the package's docs/ directory was out of place.

Contents

New page data-tests/metric-stability, registered under Other Tests beside the other threshold-style tests (execution-sla, data-freshness-sla, volume-threshold). Follows the volume-threshold page's structure.

It covers:

• What the test detects: a settled bucket's own aggregate changing after the fact, which comparing different periods does not directly check.
• min_bucket_age, and why measurements taken while a bucket is still settling are excluded from both baselines.
• last_check vs first_check, including that last_check accepts each new value automatically while first_check keeps failing until the value returns within tolerance.
• Coverage and cost: what days_back and backfill_days bound, and that the derived window is a convenience default rather than a retention policy.
• Per-dimension history, since each bucket/dimension combination is its own metric with its own baseline.
• Failure detail in test result samples, and the difference between value_changed and missing_bucket.

Also adds metric_stability to the Relevant tests line on the existing dimensions page, which the test supports.

Note on limits

The page states plainly that there is no accept-or-reset-baseline operation, that the first_check baseline is the earliest retained measurement rather than an approved snapshot, and that the test is not a substitute for an auditable financial close. Worth a look to check that framing is how you want it said publicly.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Advanced

Run ID: fa8d89e0-2e43-447f-a5c4-8582fba7c2dd

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

_{Comment @coderabbitai help to get the list of available commands.}