Fix a lag spike in the audio bus editor

[MewPurPur]

I disconnected bus_layout_changed from the function that rebuilds audio buses, as renaming audio buses causes the whole audio bus editor to rebuild which is undesired. This required a hefty workaround to avoid a lag spike when renaming, and it also caused duplicating to rebuild the audio bus editor twice just to add a new bus, and to then rename it. Now I removed the hefty workaround.

On top of unnecessary rebuilds from renaming, UndoRedo often asked for YET another rebuild. Rebuilding is an expensive operation that can cause a visible lag spike, especially if you have a lot of audio buses, which made this blunder quite noticeable. The rebuilds UndoRedo asks for are now the only ones I've kept (actually I removed one when resetting a bus volume, as you can ask for a single bus to be updated there)

• Duplicating buses: 3 rebuilds -> 1 rebuild
• Adding, moving, and deleting buses: 2 rebuilds -> 1 rebuild
• Resetting bus volume: 1 rebuild -> no rebuild

Regarding the comment I wiped - I'm positive that renaming audio buses should emit bus_layout_changed - as it changes how you interact with the audio bus system, not just how it affects audio. For example, AudioStreamPlayer relies on this signal to have an up-to-date list of buses you can pick in the inspector.

[MJacred]

For renaming audio buses, I wish for a new signal: bus_name_changed(bus_index, old_name, new_name)

For now, we could add bus_name_changed(bus_index, old_name, new_name) additionally (yes, 2 signals emitted at the same time)
and make a second PR with compatibility breaking changes, i.e. not emitting bus_layout_changed when changing the name.

My reasoning is: When I developed an audio manager plugin, any bus renamings in the editor ended up with me having to cache all bus names and then on each bus_layout_changed signal I need to re-check if (1) a rename happened and (2) guess/deduce what the new name is and (3) change bus name for all managed audio files.
At least 1 + 2 would become obsolete right away.

[MewPurPur]

Hmm okay. I don't know enough about audio to judge if this is a reasonable compat-break?

[MJacred]

Once bus_layout_changed is not emitted anymore for renames, the only other internal change is that all 3 AudioStreamPlayer* classes need to listen to that signal so that they call notify_property_list_changed();.

And if any project relies on the bus_layout_changed for bus renames, they would need to connect to the new signal. Only a simple fix would be required, but it counts as compat-breaking I reckon?

Hm… I guess I can open an issue+PR and then see what the feedback is

[MewPurPur]

Yes, please open a proposal! Would be good to bring it up with audio contribs. These systems were implemented by reduz like 7 years ago and I think I can easily implement this change if we agree on it.

(actually I'm going to make an alternative PR in a bit)

[YuriSizov]

Note on merging: This is one of two alternatives to address the problem, #81641 is the other one. We likely need to pick just one, because they don't really make sense together.

Edit: On RC reduz confirmed that #81641 is a better option.

[akien-mga]

Superseded by #81641.