Can I please request some benchmarks?

1. Making a very large (>500MB, 2GB is the limit iirc) buffer, then making a second, slightly bigger one, and destroying the original. My project allocates a lot of very large buffers, but when I need to grow them I make a bigger one, copyBuffer to transfer the contents, and then destroy the old. I'm very curious how wgpu's memory allocators handle this.
2. Bind group creation benchmarks (currently we don't cache these between frames).

1. Should already be covered by the buffer creation benchmark
2. Can add that, yeah