WIP cosmic-image-capture-source-unstable-v1 for workspace capture

[ids1024]

This will be needed to replace cosmic-screencopy-v2 with ext-image-copy-capture-v1.

This could also use ext-workspace-v1 handles, but we'll need to add that in cosmic-comp first. That could also be contributed upstream. Though on other compositors, workspaces may be on multiple monitors, so either this needs a an output argument again, or it will capture a single image for multiple outputs. (Or there could be a way to do both, if that's desired.)

[Drakulix]

I don't we have to transition everything at once, even though I would love to switch over to ext-workspace soon. But indeed it would be annoying to have to update this protocol so soon, if we want to use the ext-workspace handle either way. :/

[ids1024]

Yeah. And if we needed to support a capture source protocol for cosmic-workspace and one for ext-workspace in parallel for a while, that shouldn't be too bad. Given this is the simplest sort of protocol possible. That's more of a hassle with some other things.

But the other question is if we want to update our existing cosmic workspaces protocol to depend on ext workspace in the same way the toplevel protocol was updated in #29. That would mean not needing to replacing move_to_workspace and workspace_enter/workspace_leave with new requests/events. But workspace_enter/workspace_leave would become racy if they depend on cosmic handles created with a request rather than event (currently they're not, at least if the workspace global is bound first.)

If we want a completely new cosmic workspace extension protocol, it may be best not to add new requests to the existing one (as in #48). But maybe that's not particularly a problem either.

[Drakulix]

But the other question is if we want to update our existing cosmic workspaces protocol to depend on ext workspace in the same way the toplevel protocol was updated in #29.
...
If we want a completely new cosmic workspace extension protocol, it may be best not to add new requests to the existing one (as in #48).

Yeah I think it makes sense to ext-workspace as a base and add everything else we need in an extension protocol, rather than reworking it exactly as cosmic-toplevel works now. The latter was mostly done, because ext-toplevel-management isn't a thing yet, so we still need cosmic-toplevel for almost everything.

This is absolutely not the case with ext-workspace, where we really want to re-evaluate how to extend it, I feel.