Add RDNA4 (gfx12) support and a 0.1.2 optimization round

RDNA4 (experimental; untested on RDNA4 hardware)
- kernels/common/wmma_layout.h picks the WMMA operand/accumulator layout
  per target; K (pwin_*) and M (swin_*) index through it, so the same
  sources build for gfx11 and gfx12. D4R_WMMA_LAYOUT=12 on gfx11 is a test
  shim that runs the gfx12 indexing on gfx11 WMMA.

- M can use native e4m3 FP8 WMMA on gfx12 (D4R_NATIVE_FP8=1). Texture
  tails (enc0_tail, dec0_head) get gfx12 f16 and FP8 paths that match
  ZLUDA's gfx12 lowering.

- ZLUDA patch 0007: gfx12 WMMA helpers, optional native FP8 MMA
  (D4R_ZLUDA_WMMA_FP8_NATIVE), a layout override for testing, and an
  architecture argument for d4r_emit. Texture kernels are now compiled
  offline per target through d4r_emit (D4R_ZLUDA_EMIT) instead of on the
  local GPU.

- New d4r.ini option [Kernels] NativeFp8 (default on), wired through the
  shim and d4r_config.py. On gfx12 the bridge serves kernels/<arch>-fp8
  when it is on.

- The bridge picks the KFD GPU with the most SIMDs (so an iGPU is not
  chosen over a discrete GPU); D4R_GPU_ARCH overrides it.

- kernels/build.sh and package_release.sh build gfx1100-gfx1103 and
  gfx1200/gfx1201 (plus -fp8 variants), and D4R_BUNDLE_TEX takes
  per-target folders.

Performance (RDNA3; harness byte-identical unless noted)
- ZLUDA patch 0006: m16n8k8 f16 MMAs on WMMA (E; not bit-exact, 52-55 dB
  per layer, D4R_ZLUDA_WMMA_K8=0 disables it), per-key weight-image slots
  for native prep kernels (d4r_prep_key_at, d4r_prep_key_slots), and
  wave64 builds of MMA-free kernels (D4R_ZLUDA_WAVE64).

- M: packed FP8 encoding, conflict-free patch-merge reads, scalar weight
  loads in enc1, and weight prep once per weight set for the six enc3
  tube blocks.

- K: CU mode for the six non-wide pwin layers.

- Texture kernels: every DLSS flag variant is built natively (24 K output,
  8 M post, 4 M downsample, 4 M enc0), not only Townfall's; post and K
  output run as wave64.

- M direct output: rrlite_downsample_kernel writes the shim's output
  buffer; the shim's default direct-output presets are now 11,13.

Tools
- dump_runner takes the prep grid from d4r_prep_blocks and refuses
  texture/surface dumps.

- New pwin_check.py, swin_check.py, model_enc3.py and tail_check.hip.

Docs: README GPU table and RDNA4 notes, building (patches 0002-0007,
d4r_emit), native-kernels (RDNA4 section), performance (new numbers),
packaging README. Version 0.1.2.

packaging/D4R_README.txt

@@ -1,10 +1,10 @@
-d4r @VERSION@: NVIDIA DLSS on AMD Radeon RDNA3 under Linux and Proton
+d4r @VERSION@: NVIDIA DLSS on AMD Radeon RDNA3 and RDNA4 under Linux and Proton
 =====================================================================
 
-d4r runs NVIDIA's own DLSS Super Resolution (DLSS 4 by default) on AMD Radeon RX 7000 GPUs in
+d4r runs NVIDIA's own DLSS Super Resolution (DLSS 4 by default) on AMD Radeon RDNA3 and RDNA4 GPUs in
 DirectX 12 games under Proton. OptiScaler (included) catches the game's DLSS calls and hands them to
 d4r. d4r then runs NVIDIA's DLSS on the Radeon through ZLUDA, with the heaviest parts replaced by
-kernels written for RDNA3.
+kernels written for RDNA3 and RDNA4.
 
 This is an early release, tested on one GPU (Radeon RX 7700 XT) in about a dozen DirectX 12 games.
 SUPPORTED_GAMES.md in the source repository lists them, with per-game setup notes. Expect problems
@@ -13,9 +13,11 @@ in other games.
 
 What you need
 -------------
-- An AMD Radeon RX 7000 series GPU (RDNA3). The fast kernels are built for gfx1101 (RX 7700 XT,
-  RX 7800 XT, RX 7700). Other RDNA3 cards run DLSS without them: DLSS 3 (E) at full speed,
-  DLSS 4.5 (M) at about half speed, and DLSS 4 (K) without its transformer (worse image).
+- An AMD RDNA3 or RDNA4 GPU. This build includes native DLSS 4 and 4.5 network kernels for
+  gfx1100–gfx1103 and gfx1200–gfx1201 unless the builder selected fewer targets. Only the RX 7700 XT
+  (gfx1101) has been tested on a real GPU. gfx1201 has emulator testing; gfx1200 has compile testing
+  only. Neither establishes RDNA4 runtime support, performance or driver stability. Texture-kernel
+  optimisations are included only for targets supplied by the builder.
 - Linux with the amdgpu kernel driver (/dev/kfd). ROCm itself is not needed: the zip includes the
   ROCm 7.2.4 runtime in d4r/rocm. RocmDir in d4r/d4r.ini selects another ROCm installation instead.
 - GE-Proton 11 (tested: GE-Proton11-3), selected for the game in Steam.
@@ -58,6 +60,9 @@ Settings
 --------
 - d4r/d4r.ini: d4r's settings for this game. The main one is the DLSS model:
   K (DLSS 4, default), E (DLSS 3 CNN, a little faster) or M (DLSS 4.5, slowest).
+- On RDNA4, [Kernels] NativeFp8 is on by default. It selects native FP8 WMMA and the matching
+  gfx12-fp8 kernel folder; set it to false to use f16 widening. The FP8 path has not been tested
+  on a real RDNA4 GPU.
 - OptiScaler.ini: OptiScaler's settings, such as its fps overlay and the render resolution of each
   quality mode. Press Insert in game for OptiScaler's menu.