Diagnose: release health is empty

# Diagnose: release health is empty

## Step 1: is the release identifier set?

Release health groups by the release string on events. If the SDK sends no release (or a different one per deploy), there is nothing to aggregate. Check that `release` in init, `SENTRY_RELEASE`, or the inferred git SHA is actually populated in the running environment. Containers built without `.git` cannot infer the SHA, which silently leaves release empty.

## Step 2: are sessions being tracked?

Crash-free rates come from session data, not error events. If the SDK's session tracking is disabled or the app is a short-lived script that never produces sessions, health stays empty even with errors flowing. Long-running services (web servers, workers) produce sessions; one-shot scripts do not.

## Step 3: identifier consistency

The release on the event must equal the release you notified via the CLI. A deploy that notifies `myapp@1.2.3` while the SDK reports the git SHA splits health across two releases, both looking empty. Generate the identifier once per pipeline run and inject it into both.

## Step 4: environments

Health is per release per environment. If the release page shows the version but no health, check you are viewing the environment the events carry. An empty environment string vs `production` splits the data.

## Verify

After a deploy with a consistent identifier, open the Releases page within the hour: the version should show adoption, session counts, and crash-free percentages climbing. If sessions appear but crash-free stays blank, the session data is arriving and the aggregation is still catching up; give it time before debugging further.