Make improvements to CI/CD integration test runs (grouping, retries, health checks)

[dsawali]

With the changes introduced in #2203, we can now choose to run grouped blocks of integration tests in the CI. This may lead to potential improvements to speed or reliability.

Update the CI workflow to run groups based on subfolders and investigate the capability of doing more parallel runs to improve test runs.

Potential solutions to increase efficiency in our CI/CD runs:

• Group flakier tests into a separate running group, so that if failure occurs, the smaller group of tests can be re-run independently
• Add a health check/sanity check to ping keygen before running tests. If both endpoints (ephemeral and fresh) return 200, then continue running the tests, if it fails, cancel test run immediately

Other options:

• Add a retry method to wrap on top of existing it blocks, so that if a failure occurs in a test, it retries that block immediately