zero_init_workgroup_mem fails in some configurations

[nical]

Description
A clear and concise description of what the bug is.

On some platforms cargo nextest run fails in zero_init_workgroup_mem.

For example on a Mac M1 max:

thread 'shader::zero_init_workgroup_mem::zero_init_workgroup_mem' panicked at 'Zero-initialization of workgroup memory failed (81% of disptaches failed).', wgpu/tests/shader/zero_init_workgroup_mem.rs:175:5

Repro steps

run cargo nextest run.

Platform

• Mac M1 max (Moac OS 12.5.1 "Metal GPUFamily Apple 7") on current wgpu (rev 719ba83)
• Windows + DXC according to Update gpu allocator to 0.22 #3447 (comment)

[teoxoy]

@Elabajaba we think we fixed this issue in gfx-rs/naga#2259 (#3512).
Can you confirm that on your Windows + DXC setup it now succeeds?

[nical]

The mac M1 issue is fixed.

[Elabajaba]

Can you confirm that on your Windows + DXC setup it now succeeds?

DX12 now passes all the tests with both FXC and DXC. (apart from the 1 skipped test)

[cwfitzgerald]

Sounds like this has been fixed!