Use initializer to ensure that all worker run worker_setup()

[samtygier-stfc]

Issue

Closes #1918

Description

When creating the multiprocessing pool we want to have each work run the worker_setup() function which ensure that it is started, and that it has the operation modules loaded.

For low core counts calling pool.map() usually did the job. But it does not guarantee that each task goes to a different worker. So on machines with large numbers of CPU its possible for some workers to run the task multiple times, and others to not run it at all.

Instead use the initializer which runs a task on each worker.

"If initializer is not None then each worker process will call initializer(*initargs) when it starts." --
https://docs.python.org/3/library/multiprocessing.html#module-multiprocessing.pool

We no longer need to pass a dummy argument to setup_worker().

Testing & Acceptance Criteria

I can't reliably trigger this issue, even on the large 60-core IDAaaS workspaces.

Best is to check that it does not make things worse. I.e. check that MI starts up and an operations such as Rotate Stack work.

Documentation

Release notes

[JackEAllen]

The change seems like a reasonable solution without testing so happy to approve. Due the difficulty of trying to replicate the issue this PR is trying to fix, we should probably check this from time to time post merge to ensure we have definitely resolved the issue. We should add this as a manual test to our internet manual testing spreadsheet to check whenever we do a release for the next couple of releases..