Make MWorkers reconnecting to EventPublisher if it was killed

[vzhestkov]

What does this PR do?

It makes MWorkers to be able to recoonect with salt.transport.ipc.IPCMessageClient to the EventPublisher after killing it with OOM or any other reason, so it makes salt-master to be able to self recover after killing the EventPublisher.

Implementation of timeout and tries is a kind of bonus there, the main fix is in the way of handling StreamClosedError exceptions.

The reason why there is norelated upstream change:

salt.transport.ipc is specified as deprecated for 3009 and in the master either tcp or zmq can be used, but they have different implementation of this part which can be not affected, but it's better to check it deeper on switching the version.

What issues does this PR fix or reference?

Tracks: https://github.com/SUSE/spacewalk/issues/23526

Previous Behavior

On killing EventPublisher subprocess the salt-master is still running but is not able to handle the incoming events the proper way as the stream to the EventPublisher in MWorker was closed and not restored, so the salt-master is running but not functional.

New Behavior

The salt-master is able to self recover on killing EventPublisher and continue handling the events.

Merge requirements satisfied?

[NOTICE] Bug fixes or features added to Salt require tests.

• Docs
• Changelog - https://docs.saltproject.io/en/master/topics/development/changelog.html
• Tests written/updated

Commits signed with GPG?

Yes/No

Please review Salt's Contributing Guide for best practices.

See GitHub's page on GPG signing for more information about signing commits with GPG.