Upgrade Mesa to 24.3.4 in CI

[Vecvec]

Is your feature request related to a problem? Please describe.
The current version of Mesa contains a bug in it that causes it to give a segmentation fault if you reset query pools that get acceleration structure info. This affects #6609 because it uses a query pool to get the compacted size (only on vulkan).

Describe the solution you'd like
A newer version of Mesa to be used in CI

Describe alternatives you've considered
Just skip mesa on this test (might miss other bugs).

Additional context

[cwfitzgerald]

Kicked off https://github.com/gfx-rs/ci-build/actions/runs/12980646329

[cwfitzgerald]

Now that this is up, can you push a change that bumps the env fields in the CI .ymls

[Vecvec]

Push checks on linux give

[2025-01-30T15:38:48Z INFO  wgpu_hal::vulkan::instance] 	objects: (type: INSTANCE, hndl: 0x5603eeff2a10, name: ?)
[2025-01-30T15:38:48Z INFO  wgpu_hal::vulkan::instance] GENERAL [Loader Message (0x0)]
    	Found ICD manifest file /home/runner/work/wgpu/wgpu/icd.json, version 1.0.0
[2025-01-30T15:38:48Z INFO  wgpu_hal::vulkan::instance] 	objects: (type: INSTANCE, hndl: 0x5603eeff2a10, name: ?)
[2025-01-30T15:38:48Z DEBUG wgpu_hal::vulkan::instance] GENERAL [Loader Message (0x0)]
    	Searching for ICD drivers named /home/runner/work/wgpu/wgpu/mesa/lib/x86_64-linux-gnu/libvulkan_lvp.so
[2025-01-30T15:38:48Z DEBUG wgpu_hal::vulkan::instance] 	objects: (type: INSTANCE, hndl: 0x5603eeff2a10, name: ?)
[2025-01-30T15:38:48Z ERROR wgpu_hal::vulkan::instance] GENERAL [Loader Message (0x0)]
    	libLLVM.so.18.1: cannot open shared object file: No such file or directory
[2025-01-30T15:38:48Z ERROR wgpu_hal::vulkan::instance] 	objects: (type: INSTANCE, hndl: 0x5603eeff2a10, name: ?)
[2025-01-30T15:38:48Z ERROR wgpu_hal::vulkan::instance] GENERAL [Loader Message (0x0)]
    	loader_icd_scan: Failed loading library associated with ICD JSON /home/runner/work/wgpu/wgpu/mesa/lib/x86_64-linux-gnu/libvulkan_lvp.so. Ignoring this JSON
[2025-01-30T15:38:48Z ERROR wgpu_hal::vulkan::instance] 	objects: (type: INSTANCE, hndl: 0x5603eeff2a10, name: ?)
[2025-01-30T15:38:48Z ERROR wgpu_hal::vulkan::instance] GENERAL [Loader Message (0x0)]
    	vkCreateInstance: Found no drivers!

I'm not sure how to deal with this.

Edit: seems older version never mentions libLLVM (searched binary), they instead mentioned libLLVM-15.

[cwfitzgerald]

Derp, this it was built with 24.04, but we're using 22.04 - could you bump the version in CI, then change any instances of jammy to noble (notably on the vulkan sdk).

You can make a PR and if there are any issues I can fix it up.

[Vecvec]

Seems to work properly now!