Configure/Provision Cloud Storage bucket

[data-sync-user]

The new GCS bucket should include a lifetime policy that GCs objects with a committed=false flag after a certain TTL (24 hours?)

┆Issue is synchronized with this Jira Task

[data-sync-user]

Taddes Korris commented from Jira:

I've got the stage + prod buckets provisioned in Terraform (PR: https://github.com/mozilla/webservices-infra/pull/11481), but two things I want to confirm before merging:

1. Lifecycle policy — the ticket asks us to GC objects flagged committed=false after ~24h. GCS native lifecycle rules can't match custom object metadata (only age, storage class, etc.), so as written the rule just deletes everything older than a day. Do we want a dedicated bucket/prefix where committed objects get moved out (so only stale-uncommitted ones age out), or app/job-side cleanup that actually reads the committed flag? Want to make sure I implement the right mechanism.
2. Bucket name — I went with payloads (resolves to sync-prod-prod-payloads / sync-nonprod-stage-payloads). Does that match what syncstorage-rs will reference, or should it be something else?

Will adjust once we land on these.

[data-sync-user]

Taddes Korris commented from Jira:

Spun up a dev-only copy of the bucket setup so we can test it in dev first before stage/prod land: https://github.com/mozilla/webservices-infra/pull/11491

Same infra as the stage/prod PR (https://github.com/mozilla/webservices-infra/pull/11481), just scoped to the dev env — bucket resolves to sync-nonprod-dev-payloads. The same two open questions above (lifecycle mechanism + bucket name) apply to both.