test(shader): cover integer image atomics

[asimplestray]

What

Test-only follow-up: the 6 integer image atomics (IMAGE_ATOMIC_CMPSWAP/SUB/SMIN/SMAX/INC/DEC) have since landed in main via other work with full SPIR-V backend plus a Vulkan execution test, so this PR keeps only the decode-level regression test.

Adds the image_atomics test (decode opcode + family + image classification plus translation emitting the expected IR op for all 6 encodings, including the dmask=3 requirement for cmpswap).

Tested

Linux (GCC, BUILD_TESTING=ON, ANYPS5_ENABLE_SPIRV_TOOLS=OFF): full build, new test passes. Note: agc_driver_pixel_helper_lanes fails on current main independent of this PR (Vulkan pixel-helper assertion, no shader implementation touched here).

Checklist

• Based on current main; no other open PR adds the same test (checked open shader PRs)
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (decoding unknown opcodes still throws)
• No notes, investigation .md files or images added to the repository (attach them to this PR)
• New third-party code is a submodule built from source (none added)
• Generic behaviour, not specific to one title
• Depends on: none
• AI-assisted: yes

[github-actions]

Other open pull requests touch the same code. Merging one will break the other, or both implement the same thing. If this one builds on them, list them in Depends on.

Pull request	Conflicting files (lines on main)	Same exports	Same new files
#295	core/libs/prx/libSceAgcDriver/CMakeLists.txt L806
#375	core/libs/prx/libSceAgcDriver/CMakeLists.txt L806
#423	core/libs/prx/libSceAgcDriver/CMakeLists.txt L806
#448	core/libs/prx/libSceAgcDriver/CMakeLists.txt L806

[oneandonlydean]

Thanks for this. core/libs/prx/libSceAgcDriver/tests/execution/ImageAtomics.cpp on main already runs all six of these integer image atomics (sub, smin, smax, inc, dec, cmpswap) on the GPU and checks the results, and #667 added the bitwise and unsigned ones next to it. That covers decoding and translation too, so a separate decode-level test doesn't add coverage, and I think this can be closed. If you see a case the execution tests miss (another dmask or a 64-bit form, for example), adding it there would be welcome.

[oneandonlydean]

As I mentioned on 10-06, tests/execution/ImageAtomics.cpp on main already runs all six of these integer image atomics on the GPU and checks the results, and #667 added the bitwise and unsigned ones. So I'm closing this as superseded. Thanks for the work.

If you find a case those tests miss (another dmask or a 64-bit form, for example), a PR adding it there is welcome. If you think I've missed something here, please reopen.