WebGPURenderer: Introduce indirect drawing support

[sunag]

Related issue: #29594

Description

This PR introduces indirect drawing support and a simple example in webgpu_particles.
Part of the work and also related to this result in #29594 PR

const uint32 = new Uint32Array( 5 );
uint32[ 0 ] = indexCount;	// indexCount
uint32[ 1 ] = fireCount; 	// instanceCount
uint32[ 2 ] = 0;		// firstIndex
uint32[ 3 ] = 0;		// baseVertex
uint32[ 4 ] = 0;		// firstInstance

const indirectAttribute = new THREE.IndirectStorageBufferAttribute( uint32, 5 );

geometry.setIndirect( indirectAttribute );

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

	Before	After	Diff
WebGL	689.82 170.9	689.92 170.93	+97 B +33 B
WebGPU	812.95 218.95	813.52 219.09	+572 B +140 B
WebGPU Nodes	812.46 218.82	813.03 218.94	+572 B +123 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

	Before	After	Diff
WebGL	462.84 111.78	462.94 111.81	+97 B +27 B
WebGPU	536.92 144.9	537.35 145	+433 B +98 B
WebGPU Nodes	493.03 134.63	493.46 134.72	+433 B +96 B

[Spiri0]

I thought I would use another interesting example to test this and it works really well

https://codepen.io/Spiri0/pen/zYgzgJR?editors=0111

When I wanted to test this in my app, it kept giving me this error message over and over again

The console messages drawIndexedIndirect and draw come from my analysis. After hours of troubleshooting, I found the cause and it can easily be reproduced using this example. The cause is "material.wireframe = true" which I always had on in my app and that forces the indexed path in the backend. Everything is fine without material.wireframe = true;
So this is not an error in the drawIndirect logic.

Now I need a break, it was a Sisyphean task to find that. I could look into this over the next week as to why wireframe = true forces the indexed path even if you don't set any indexes in InstancedBufferGeometry