config: workspace empty behavior

[kotarac]

Draft until I manage to go about testing this on multiple outputs because of the workspace restoring logic.

Implementation according to our discussion here #541.
Closes #519.

[mahkoh]

In case you wrote the book changes by hand: You don't have to do that. There is already enough busy work when adding new features.

[kotarac]

In case you wrote the book changes by hand: You don't have to do that. There are already enough busy work when adding new features.

I did not actually. I've had gpt-5.2 write the book content and before that help me better clarify the behavior of each value in spec.yaml and review the changes. Or to be precise, review the changes and then do all that.

Btw, possibly interesting piece of info: I've tried letting it help write a testcase in the integration test and I ended up with a couple GiB worth of logs in testruns/ dir because it destroyed a wl_surface before destroying a wl_viewport, resulting in leak tracking going crazy with logging.

Still need to get around to testing this on multiple outputs. I only have such setup at the office and unfortunately I haven't had any time lately to sit down and go through it.

[kotarac]

In case you wrote the book changes by hand: You don't have to do that. There is already enough busy work when adding new features.

Also, fwiw, while this is true and the book and control center added more, I don't really mind this at all.

In my 1st Jay contribution on the touchpad stuff, it took quite some time to go through everything, but the 1st moment I've had it compile it actually worked correctly (despite having multiple style/consistency issues) which is at least partly because of the way it's set up right now.

I suppose some of it could be distilled now to somehow have 1 central place which dictates how all "touchpoints" work, but it's not exactly a big deal once you know where everything is.