Add EventsPipeline and testing utilities

[jakekaplan]

Implement the EventsPipeline and the events_pipeline fixture.

With client side task run orchestration, task-runs are now eventually consistent. There are a lot of places in our tests where we are reading task runs back from the server to assert something about that finished task run. We still want to support that pattern. To do this, this PR introduces the following:

async def test_run_task(prefect_client: PrefectClient, events_pipeline):
    @task
    async def foo():
       pass
       
    # run the task  
    await run_task_async(foo, task_run_id=task_run_id) 
    
    # run the events pipeline to ensure eventual consistency is actually achieved
    await events_pipeline.process_events()

    task_run = await prefect_client.read_task_run(task_run_id)
    assert task_run.id == task_run_id

Now you can run the task in your test. And then manually execute a version of the events processes that would normally happen on your sever. In the future EventsPipeline can also be used at teardown time of the ephemeral api, so that events and tasks can be recorded.

[jakekaplan]

In the future when shutting down the ephemeral version we can invoke this method on everything sitting inside of the in-memory events queue

[codspeed-hq]

CodSpeed Performance Report

Merging #14765 will not alter performance

_{Comparing run-events-pipeline (115f6ee) with main (b7cca3d)}

Summary

✅ 5 untouched benchmarks

[cicdw]

minor nits, makes sense to me