Java SDK CIs Improvement Tracking Ticket

Problems:
1. GAPIC Unit Tests Long Duration

On changes that impact all of google-cloud-java, all the GAPIC unit tests are run. These test jobs take well over 30+ min (https://github.com/googleapis/google-cloud-java/actions/runs/26462248492/job/77913201176?pr=13222). One latest log results timed this to be ~38 min.

These GAPIC unit tests are autogenerated within each module and do not vary too much. These look to test happy and sad paths for each RPC method, but generally are the same tests with different inputs and outputs. We have dedicated showcase tests that can test these RPCs functionality once and ensure that the common variations are tested properly.

Question(s):
- Can we stop autogenerating these unit tests and instead rely on showcase unit tests for complete coverage? Additionally, we can rely one showcase ITs to ensure that the functionality is able to work E2E

2. GraalVM Tests Duration

The GraalVM jobs (A,B,C) run through test cases (units and ITs) in each module twice (once to investigate all the possible code paths and a second time to run through the tests). This doubles the amount of time needed to ensure GraalVM compatibility. One of the latest runs for GraalVM ran for ~90 min (https://btx.cloud.google.com/invocations/a827908d-a2fb-49f2-8c39-01b51e4baad8/targets).

Is there a way that we can streamline the tests to run faster?
- Remove the GAPIC unit tests so that they do not need to run? Instead, we rely on showcase unit tests and integration tests
- Only run the required tests once?
- Use GraalVM fast build mode?

2.1 GraalVM Tests Flakiness

Additionally, the test suite is already quite flaky and this flakiness compounds into an even more expensive operation when there are long test times.

3. Enforcer and Fmt Jobs Duration

Similar to issues above, running these jobs is expensive and take a long time if they require building and compiling all the GAPIC modules.