Make integrations tests more stable

[vdusek]

Our integration tests are highly unstable, often requiring multiple attempts to pass. Below are the two main issues we are encountering...

Issues during Actor creation

The first problem occurs when creating an Actor in conftest/make_actor (conftest.py#L247):

    ...
    build = await actor_client.build(version_number='0.0', wait_for_finish=300)
    assert build['status'] == ActorJobStatus.SUCCEEDED

The test frequently fails with the following assertion error:

>       assert build['status'] == ActorJobStatus.SUCCEEDED
E       AssertionError: assert 'READY' == <ActorJobStat...: 'SUCCEEDED'>
E         
E         - SUCCEEDED
E         + READY

Issues during Actor execution

Another common issue arises when checking the Actor's run status in the integration tests:

    ...
    run_result = await actor.call()
    assert run_result is not None
    assert run_result['status'] == 'SUCCEEDED'

Here, the test often fails with a similar mismatch in the expected status:

>       assert build['status'] == ActorJobStatus.SUCCEEDED
E       AssertionError: assert 'READY' == <ActorJobStat...: 'SUCCEEDED'>
E         
E         - SUCCEEDED
E         + READY

[vdusek]

See the timeout=60 in https://docs.apify.com/api/v2#tag/ActorsRun-collection/operation/act_runs_post ... it could be related.

From Apify client:

wait_for_finish: The maximum number of seconds the server waits for the build to finish before returning. 
    By default it is 0, the maximum value is 60.

[vdusek]

Revert changes in #306 once this is done.