GeezerVison

precise-code-intel-worker and syntactic-code-intel-worker never got the chart's Redis settings, so they always connected to the default redis-store:6379. With external Redis that host doesn't exist. Both workers use Redis for a periodic lock around their queue metrics.

This adds the same sourcegraph.redisConnection env vars and checksum/redis annotation that frontend, gitserver, searcher, and worker already use.

If you set REDIS_*_ENDPOINT in preciseCodeIntel.env or syntacticCodeIntel.env as a workaround, remove it when upgrading.

Checklist

• Follow the manual testing process
• Update changelog
• Update Kubernetes update doc

Test plan

• helm unittest: the Redis connection tests now cover both workers, including existingSecret. They fail without the template changes.
• helm template with external Redis secrets renders both env vars and the annotation on each worker.
• On a live instance with external Redis, the precise worker logged lookup redis-store errors. After the env vars were added by hand to the syntactic worker, traces showed it taking its Redis lock. Not deployed from this branch.