Add Heightfield mask to GPUParticlesCollisionHeightField3D

[Rudolph-B]

Fixes #101750

As described in the issue, particles would collide with the icon of any node within the GPUParticlesCollisionHeightField3D AABB. Additionally, I noticed that an object's layer mask had no effect on whether it was considered by the height field.

This change addresses both issues.

[Rudolph-B]

@clayjohn, I changed to comparing layer_mask against cull_musk . To accomplish this I needed to do a few extra things:

• I needed to add functions to particle_storage to pull the cull_mask (I'm not really confident I followed standard here)
• I needed to change the default value of the cull_mask to (1 << 20) - 1 (this excludes the gizmo and other similar layers)

[clayjohn]

@clayjohn, I changed to comparing layer_mask against cull_musk . To accomplish this I needed to do a few extra things:

• I needed to add functions to particle_storage to pull the cull_mask (I'm not really confident I followed standard here)
• I needed to change the default value of the cull_mask to (1 << 20) - 1 (this excludes the gizmo and other similar layers)

You shouldn't change the default value to support this. You would be potentially breaking users' games in order to make something easier in the editor.

[Rudolph-B]

I have removed the default value change.

I’m not entirely sure what happened. At some point during my testing, it was necessary to change the default value, but when I started testing again today, it was no longer required to fix the bug.

[Calinou]

I can still reproduce the issue in the MRP with the latest revision of PR:

[Rudolph-B]

I see, I just reloaded the stored cull_mask in the project instead of the default one.

[Rudolph-B]

As discussed in the rendering meeting I pivoted to adding a specific mask to GPUParticlesCollisionHeightField3D. This allows us not to break compatibility and to better match expected behavior.

I added @yahiaetman as a co-auther since I based this of their work in #93291

[clayjohn]

Just some fixes for the documentation. The code looks great!

I wasn't able to reproduce the MRP, but I trust that you have tested carefully and can confirm that it is working properly

[Repiteo]

Thanks!