rr: add FSR Ray Regeneration, exact RDNA3 inference and motion experiments

Add an experimental FSR Ray Regeneration backend for D3D12 RR requests:
- Convert depth, motion, normals and albedos into AMD denoiser inputs.
- Split and recompose diffuse/specular signals, then run Super Sampling.
- Add stability/detail controls, demodulation guards and halo limiting.
- Handle AMD provider discovery under NVIDIA GPU spoofing.
- Add runtime probes, synthetic reference scenes and per-dispatch profiling.

Add exact gfx11 DLSS-RR inference using f16 cooperative matrices:
- Reproduce FP8 conversions, accumulation and camera reprojection arithmetic.
- Widen model weights and avoid requiring VK_EXT_shader_float8.
- Extend model building, replay tools and descriptor metadata.

Expand motion preprocessing and experimental AW2 native-motion support:
- Add tiled LDS dilation, padded allocation handling and thin-only sky fill.
- Order dilation with input waits and NGX on the selected stream.
- Add jitter-aligned native composition, hybrid routing and wire history repair.
- Add prepared masks, sparse repair and independent motion/repair controls.

Add optional fused R11G11B10 input/output conversion, sampled stage timings
and profile summarization. Extend configuration, packaging, documentation,
FidelityFX headers and CPU/GPU regression coverage.

tests/test_profile_summary.py

@@ -0,0 +1,48 @@
+"""Keep startup, feature instances and resolution changes out of shared medians."""
+import importlib.util
+from pathlib import Path
+import unittest
+
+ROOT = Path(__file__).resolve().parents[1]
+spec = importlib.util.spec_from_file_location('profile_summary', ROOT/'scripts/summarize_d4r_profile.py')
+summary = importlib.util.module_from_spec(spec)
+spec.loader.exec_module(summary)
+
+
+class ProfileSummaryTests(unittest.TestCase):
+    def test_warm_groups_and_unavailable_timings(self):
+        def sample(frame, render='853x480', reset=0, gpu='2.2'):
+            return (f'D4R_PROFILE frame={frame} render={render} output=2560x1440 '
+                    f'preset=11 quality=3 flags=0x43 input_wait=cpu reset={reset} '
+                    f'vk_total=2.8 dlss_gpu={gpu} native_motion_gpu=-1 wire_repair_gpu=-1')
+        create='NVSDK_NGX_D3D12_CreateFeature(feature=1)'
+        log='\n'.join((create, sample(1), sample(300, reset=1), sample(600),
+                       sample(900, gpu='nan'), sample(1200, render='1706x960'),
+                       create, sample(300)))
+        rows=summary.summarize(log)
+        self.assertEqual(len(rows), 3)
+        self.assertEqual([r['samples'] for r in rows], [2, 1, 1])
+        self.assertEqual([r['feature_instance'] for r in rows], [1, 1, 2])
+        self.assertEqual(rows[0]['timings_ms']['dlss_gpu']['n'], 1)
+        self.assertIsNone(rows[0]['timings_ms']['wire_repair_gpu'])
+        self.assertEqual(rows[0]['timings_ms']['vk_total']['median'], 2.8)
+
+    def test_routing_changes_get_separate_groups(self):
+        base = ('D4R_PROFILE frame=300 render=853x480 output=2560x1440 '
+                'preset=11 quality=3 flags=0x43 input_wait=cpu reset=0 '
+                'feature=29184 native_field=1 wire_requested=1 gpu_completion=0 ')
+        rows = summary.summarize(base+'native_substitute=0\n'+base+'native_substitute=1')
+        self.assertEqual(len(rows), 2)
+        self.assertEqual([r['native_substitute'] for r in rows], ['0', '1'])
+
+    def test_scene_window_and_actual_output_mode(self):
+        base = ('D4R_PROFILE frame=600 render=853x480 output=2560x1440 '
+                'preset=11 quality=3 flags=0x43 input_wait=cpu reset=0 ')
+        log = ('[ 10.000] '+base+'vk_total=8 vk_output_mode=transfer\n'
+               '[ 20.000] '+base+'vk_total=3 vk_output_mode=compute\n'
+               '[ 21.000] '+base+'vk_total=4 vk_output_mode=transfer\n'
+               '[ 30.000] '+base+'vk_total=9 vk_output_mode=compute')
+        rows = summary.summarize(log, since_clock=19, until_clock=22)
+        self.assertEqual(len(rows), 2)
+        self.assertEqual([r['vk_output_mode'] for r in rows], ['compute', 'transfer'])
+        self.assertEqual([r['timings_ms']['vk_total']['median'] for r in rows], [3, 4])