[Task]: Clean up Java Example PreCommit/PostCommit Dataflow

### What needs to happen?

Currently we have multiple Example PreCommit/PostCommit on Dataflow:

- PreCommit Example V1 (8, 11, 17) - run wordcount
- PostCommit Example V1 (11, 17) - run wordcount and impersonate wordcount
- PostCommit Example V2(8, 11, 17) - run full example (except wordcount)
- PostCommit V1 - run full example (except wordcount)
- PostCommit V2 - run full example (except wordcount)

There are a few issues
- No Dataflow Pipeline run on runner v2 at PreCommit - this has caused regression escaped e.g. #28274
- Wordcount never runs on runner v2 - either on Example PostCommit or the "big" PostCommit
- Duplicate test run, e.g. both PostCommit V2 and PostCommit Example V2 runs Java example
- Example tests are defined in two different files (`runner/google-cloud-dataflow-java/build.gradle` (for runner v1 tests) and `runner/google-cloud-dataflow-java/example/build.gradle` (for runner v2 tests))

We should put all example tests to `example/build.gradle` also

### Issue Priority

Priority: 2 (default / most normal work should be filed as P2)

### Issue Components

- [ ] Component: Python SDK
- [X] Component: Java SDK
- [ ] Component: Go SDK
- [ ] Component: Typescript SDK
- [ ] Component: IO connector
- [ ] Component: Beam examples
- [ ] Component: Beam playground
- [ ] Component: Beam katas
- [ ] Component: Website
- [ ] Component: Spark Runner
- [ ] Component: Flink Runner
- [ ] Component: Samza Runner
- [ ] Component: Twister2 Runner
- [ ] Component: Hazelcast Jet Runner
- [ ] Component: Google Cloud Dataflow Runner