[Prism] Refactor stageState to a behavior interface to reduce branch combinatorics

[lostluck]

A basic no-behavior change refactoring to split the different stage kinds up more cleanly into the 3 types: ordinary, Stateful, and aggregation.

While their differences don't come up everywhere, they in particular affect when building bundles, and adding pending input elements to their stages.

• Ordinary stages are the simplest, and handle most simple batch stages.
• Stateful stages handle Stateful transforms, with state, timers, and a requirement for keys. They don't respect triggers.
• Aggregations are Stateful stages that don't handle timers, have specific conditions for triggering, as well as handling late data.

The hope of this change is to reduce the combinatorics complexity parts of prism were suffering under, depending on which kind of stage was being handled. This leads to some duplication between Stateful and aggregation stages, but having clear handling of their differences should make up for it.

This is a minimal change. Not every potential portion is moved over right now, just those that had additional branching paths due to the old Stateful or aggregate booleans. Future changes should move handling of the pending elements stacks into similar methods instead of simply checking both.

This needed to work a change that was made to ElementManager.FailBundle which tried to avoid an issue with a deadlock in one of the test stream tests (testMultiStage). The deadlock problem became much worse for reasons not evident to the split, and it appears it was due to never performing the cleanup that Fail bundle performs.

Part of prism work: #29650 and was originally mentioned in #33881

---

Thank you for your contribution! Follow this checklist to help us incorporate your contribution quickly and easily:

• Mention the appropriate issue in your description (for example: addresses #123), if applicable. This will automatically add a link to the pull request in the issue. If you would like the issue to automatically close on merging the pull request, comment fixes #<ISSUE NUMBER> instead.
• Update CHANGES.md with noteworthy changes.
• If this contribution is large, please file an Apache Individual Contributor License Agreement.

See the Contributor Guide for more tips on how to make review process smoother.

To check the build health, please visit https://github.com/apache/beam/blob/master/.test-infra/BUILD_STATUS.md

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.

[github-actions]

Checks are failing. Will not request review until checks are succeeding. If you'd like to override that behavior, comment assign set of reviewers

[lostluck]

Alas, when I synced before starting this, my local builds stopped working, so I'm unable to dig into what I messed up in the migration (likely in the startEventTimeBundle handling). I'll be on a trip for a week, so I'll revisit after that.

[derrickaw]

Hi @lostluck, friendly reminder to revisit this PR. Thanks.

[lostluck]

Damon helped unblock my local testing against the Java suite. The Java build wasn't working after I synced, so I needed the reminder that beam's Gradle Clean doesn't work properly, and sometimes you just need to manually remove all the build folders: find . | grep \/build$ | xargs -I{} rm -rf {}.

[lostluck]

assign set of reviewers

[github-actions]

Assigning reviewers. If you would like to opt out of this review, comment assign to next reviewer:

R: @jrmccluskey for label go.

Available commands:

• stop reviewer notifications - opt out of the automated review tooling
• remind me after tests pass - tag the comment author after tests pass
• waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

[jrmccluskey]

LGTM - gave this a quick pass since it's mostly re-arranging existing code, the end result seems far more readable