The cloud-init status check in `pgbouncer-health-check` may not retry

https://github.com/developmentseed/eoapi-cdk/blob/main/lib/database/lambda/pgbouncer-health-check/health-check.sh#L14

I believe that there's a chance the loop in this status check doesn't work properly. If the status returns WAITING with a status of 2 then `set -euo pipefail` might fail this. I have a screenshot of SSM trying to run this command, then exiting with no message with a status of 2. The only explanation I could think of is exiting from the `cloud-init` status check.

A possible solution is to let `cloud-init` do the wait interally (`--wait`).