emu/debug: Allowed device_state_interface to participate in global symtable state

[holub]

After this, for device which implements device_state_interface we can access state from global symtable.
CPU (exec) devices continue to publish their state to separate CPU symtables (child of GLOBAL).

Also suggesting to rename machine().debugger().cpu() to machine().debugger().state() as "cpu" confusing term in this context as far as we can have multiple cpu. <<< will do separately, if requested

[holub]

I'd like those to be delivered/moved forward: #14908 #14898 #14765 #14754 #14740 #14739 #14328 #13122
Who can I blackmail for this?

[galibert]

I don't have a problem with renaming state() to cpu(), but I do have a problem with not renaming debugger_cpu to debugger_state then. And the filename too.

[holub]

Yeah, I was waiting for feedback on this change.
So... revert the field or rename the class as well?

[holub]

I made an executive decision and only kept functional changes in this PR.
If anybody support "rename", I can bring it as a separate change.

[galibert]

For other things I want to do I'm going to end up massively reorganizing that crap, so tabling the rename is a good thing

[cuavas]

So what happens when two devices in a system have a state item with the same name?

[holub]

This is unpleasant situation where last registered win. But we are far away from this situation as the device needs to be "state" and not be executable. I believe now number of devices which match this condition are close to 0.