RFC: Call services directly

[balloob]

Description:

We used to call services via the event bus. This was done to be more event driven. This also allowed back in the day to get multiple Home Assistants working together. This functionality was never fully implemented and has since been deprecated and removed.

This PR aims to simplify the service calling code. We still distinguish between blocking and non-blocking, but now exceptions are able to propagate up. This is especially important for tracking incidents with the unauthorized exception.

This is an RFC. I mainly wanted to see how fast it would be to get done… and it was fairly simple. It's a backwards compatible change.

Breaking change: The service_executed event is no longer being fired when a service is done.

Checklist:

• The code change is tested and works locally.
• Local tests pass with tox. Your PR cannot be merged unless tests pass
• There is no commented out code in this PR.

If the code does not interact with devices:

• Tests have been added to verify that the new code works.

[balloob]

I will not fix the tests until I got some feedback about this idea.

[MartinHjelmare]

What event info are we loosing with this? There's still an event fired after executing the service, right?

[balloob]

So we will still fire an event that we're executing a service and that shares the context with all following changes.

The only event that is missing is "service executed", which always only contained an ID to map it back to the blocked service call.

The thing we're getting is that for blocking calls we get direct feedback that the service doesn't exist or if the data is invalid, which we can now communicate back to the caller. The timeout eventually can also move out of the core and should really be controlled by the caller.