I believe this is unnecessary in CI context, since tests will still fail if the database is unavailable, and since the database is very unlikely to fail transiently. None of the 2017 database instances have even once rebooted in a year (as measured by the fact their non-permanent, non-EIP IP addresses have ever changed). No evidence that they have ever gone unavailable for short or long periods either.
Right - we only need to wait in the development environment. And if you remove the wait script, one of the packages is also unnecessary - the one that carries psql ;-).
I believe this is unnecessary in CI context, since tests will still fail if the database is unavailable, and since the database is very unlikely to fail transiently. None of the 2017 database instances have even once rebooted in a year (as measured by the fact their non-permanent, non-EIP IP addresses have ever changed). No evidence that they have ever gone unavailable for short or long periods either.