Use Vulkan Memory Allocator 3.0.1 and switch to Vulkan 1.3 for continuous integration

[IAmNotHanni]

Closes https://github.com/inexorgame/vulkan-renderer/issues/458 For more information read VMA's release note for 3.0.0 and release note for 3.0.1

Changes

• [x] Use vulkan-memory-allocator/3.0.1 version in conanfile.py
• [x] Remove deprecated Vulkan Memory Allocator record&replay feature
• [x] Use new VMA memory allocation naming system (vmaSetAllocationName)
• [x] Update VMA dump file visualization tool to version 3.0.0
• [x] Remove deprecated VMA memory budget calculation
• [x] Swich to Vulkan 1.3 in the CI

Notes

• I removed the old VMA memory statistics calculation at the exit of the program, since it's not that useful for now
• I am still using the old memory creation flags like VMA_MEMORY_USAGE_GPU_ONLY. They are still supported for compatibility, read VMA's release note for 3.0.0.

Testing

It would be nice if @westernheld and @aschaeffer could compile and run the code to also verify nothing is broken :)

[IAmNotHanni]

Broken CI build

I guess something with the way we configure VMA is wrong. There were some changes: https://github.com/GPUOpen-LibrariesAndSDKs/VulkanMemoryAllocator/releases/tag/v3.0.0

Solution

Ah it's much simpler, we need to switch to Vulkan 1.3 for this probably: https://github.com/inexorgame/vulkan-renderer/blob/main/.github/workflows/build.yml

env:
  inexor_build_type: "Release"
  inexor_vulkan_version: "1.2.189.0"
  inexor_vulkan_sdk: "$GITHUB_WORKSPACE/../vulkan_sdk/"

[IAmNotHanni]

Yes, that fixed it. Opinions?