fix(agc): skip prim state preparation for a patch/hull mismatch

[MichonGoddijn231849]

What

Since #639, sceAgcCreatePrimState also resolves the graphics ABI ahead of time, and that resolve throws when the primitive type is a patch list without HS_EN, or a hull stage is set with another primitive type. GTA V (PPSA04264) creates such a prim state while it loads, so on current main it aborts a few seconds after launch with AGC graphics: VGT_SHADER_STAGES_EN=0x2002000: Patch topology and HS_EN disagree.

No draw can use a state like that, so the ahead-of-time preparation is skipped for it and the packed registers are still returned. A draw that does use such a state still fails in the draw-time check, as before. Bit 2 of VGT_SHADER_STAGES_EN is HS_EN, not an NGG bit, so the constant is renamed.

The fix and its test are by @CouchRiv, who prepared it on Linux and asked me to open it.

Tested

Windows 11 Pro, AMD Ryzen 5 7600, NVIDIA GeForce RTX 5080, MinGW-w64 GCC 15.2.0, on main d70b899.

• agc_shader_helpers (with the new case) and guest_shader_link pass.
• GTA V (PPSA04264): no longer stops on the patch/HS_EN mismatch. It then stops at the next registration-time error, shader image heap capacity exceeded (Sampled image heap of 16 rejects a retail shader before the metadata limit #1789).
• @CouchRiv: the new case fails on main and passes with the change; full ctest 516/516 on RADV and lavapipe.

Checklist

• Based on current main; no other open PR implements the same functions
• One topic per PR; follow-ups go in a new PR
• No comments in code except technical debt
• Unimplemented paths throw (NotImplemented_nid_no_patch); silent stubs are listed in TechnicalDebt
• No notes, investigation .md files or images added to the repository
• New third-party code is a submodule built from source
• Generic behaviour, not specific to one title
• Depends on: none
• AI-assisted: yes. Written with Claude (Anthropic); I reviewed and tested it.

🤖 Generated with Claude Code

[MarcZX10]

I updated PR #2066 onto current main and moved its regression cases into a dedicated helper, so the test additions no longer share the same insertion hunk. #2066 covers optional geometry output register handling; #2293 covers the patch/hull mismatch path. They are separate fixes with no dependency.