Try GHA retry

[paulb777]

This validates a proof of concept of using GHA retry (#11253) instead of travis retry.

It doesn't address the Storage Integration flake, but it successfully retries it.

#no-changelog

[andrewheard]

LGTM, I see the 3 attempts in the logs.

[ncooke3]

LGTM. The retry_wait_seconds seems like an interesting option that we may able to leverage for cases where an instant retry may lead to the same failure as before (e.g. too many requests errors).

[paulb777]

I addedretry_wait_seconds and it successfully self deflaked. It would be nice to add a longer time only on nightly runs for the git connectivity flakes.