Prevent mapping areas with invalid IDs for Area2D/3D

[AThousandShips]

The inout code of area assumes a valid ObjectID is attached to all bodies and areas, leading to it mapping the invalid one to one entry, which doesn't make sense and creates strange behaviour

Added a separate commit for emitting the signals, uncertain if they should or should not be emitted in these cases, there's also an inconsistency between bodies between 2D and 3D which I matched there, can omit the signals if desired

Will also look into the same for bodies, and 3.x

• Fixes: Node-based Area3Ds can't tell when overlapping with PhysicsServer instanced ones. #79458

[Ughuuu]

[b]Note:[/b] [Area2D]s and [PhysicalBody2D]s created..

Also, is this correct? This makes it sound like creating them from PhysicsServer would result in instability. Is this only at the very beginning of simulation or?

[AThousandShips]

This is correct, they are not associated with a node so they work differently,

[golfinq]

Maybe add a note to the physics servers docs as well?

[AThousandShips]

This is specifically about behaviour of Area2D/3D and not the server ones, so I think it isn't relevant in the server as the areas created three don't work the same way, and don't have these signals etc. and therefore unless you read the documentation for the nodes you won't really care about those parts

The server side description is lacking but IMO outside the scope of this fix

[golfinq]

Makes sense, thanks for explaining

[YuriSizov]

I think you can squash commits, emitting here seems to follow the established logic. That logic is kind of weird and indeed can lead to the signal being emitted with some garbage object reference, but I don't think your changes are going to make it worse in any way.

[YuriSizov]

Makes sense to me

[YuriSizov]

Thanks!

[AThousandShips]

Thank you!