Investigate flaky tests related to PFS-updates

[ezdac]

During #7162 we discovered that the integration tests related to sending PFS-updates were flaky:

https://app.circleci.com/pipelines/github/raiden-network/raiden/11997/workflows/db4fb64c-fe35-4a98-ab96-e4c53fb55802/jobs/197868/tests#failed-test-0

https://app.circleci.com/pipelines/github/raiden-network/raiden/12007/workflows/dd43e248-f81e-466c-955e-401a8727e876/jobs/198232/tests#failed-test-0

• here it seems like the apps have not queued up all expected PFS-updates to the broadcast queue when the fixtures for the test yield.

Eventually this might be just a synchronisation issue where we have to wait longer in the fixture setup until all
nodes are in the state we expect in the tests?

Note that there were some changes to the synchronicity of I/O around the state machine.