CPUParticles2D - Add ability to follow physics interpolated target

[rburing]

Allows a non-interpolated particle system to closely follow an interpolated target without tracking ahead of the target, by performing fixed timestep interpolation on the particle system global transform, and using this for emission.

Ports relevant parts from 3e19cf8 and 723632a and a117a33 by @lawnjelly

Test project from #92006 (comment): particles_2d_example.zip (CPU particles are red).

• Fixes one half of Physics interpolation breaks GPUParticles2D and CPUParticles2D movement #92006

@lawnjelly Is removing the old NOTIFICATION_TRANSFORM_CHANGED really correct?

[lawnjelly]

@lawnjelly Is removing the old NOTIFICATION_TRANSFORM_CHANGED really correct?

From memory, it could well be, as the particle transforms should now be in true global space and therefore have to be recalculated each frame anyway from the global_transform.

The old particles were in some hacked global space by applying the inverse of the node transform both client and GPU side, for no sensible reason (as far as I can see, it was probably to avoid doing the changes to properly support global space, which we have since done). That hacked global space made interpolation nigh on impossible, hence the reason for introducing true global space capability for canvas items.

This may be depending on the changes to support set_canvas_item_use_identity_transform().

[lawnjelly]

Seems to work great, I tested:

• physics interpolation on, global mode
• physics interpolation on, local mode
• physics interpolation off, global mode
• physics interpolation off, local mode

I've also tested children of the CPUParticles and their transform appears to be correct too.

It is a while since I wrote this though (year and a bit) so it's not super fresh in mind, so the more testing it can get (particularly in real projects, which I don't have for 4.x) the better. Probably a good idea to try in all the relevant 2D demo projects.

[lawnjelly]

Note: This is assuming single threaded pattern for doing the 2D render, that there's no multithreaded rendering and contention for this variable going on. It's safe in 3.x, and I assume in 4.x, but worth mentioning.

(If it was, it would be solvable with something uglier).

[lawnjelly]

Looks fine as far as I can see but could do with a secondary review if any of 4.x team is up to check it.

[Calinou]

Tested locally, it works as expected. Code looks good to me.

I've tested the MRP from OP as well as the 2D particles demo with particles converted to CPUParticles2D.

With 10 physics ticks per second and interpolation enabled, this is how it looks with GPUParticles2D (which this PR doesn't fix):

gpu_particles_2d_interpolation.mp4

And with CPUParticles2D:

cpu_particles_2d_interpolation.mp4

[Repiteo]

Thanks!