Offer three denoise networks and a method to choose between them

AI Denoise's Apply switch becomes Method: Bilinear, Fast, Medium, Best,
default Best, so an untouched raw writes nothing and develops through the
mixture. `apply` is still read and never written: 0 is Bilinear, 1 keeps
a network already chosen.

- Best is the mixture of a flat and an edge expert with a learned gate;
  Medium and Fast are students distilled from it. 2.48 s, 0.79 s and
  0.57 s for a 20 MP frame on TensorRT fp16.
- Each network carries its own tile border (256 for the mixture, 192 for
  the students) through `dr_denoise::Shipped` and `TileNet::halo`.
- The file is hashed once at open and each network keys its own cached
  result; Bilinear keeps the result in memory for the way back.
- Each has an .a16w16 sibling for the Hexagon: 0.00 dB on the 6D gate,
  at most 0.11 dB with the noise scaled x0.5 to x4.
- APK BUNDLED 19 -> 23; the PKGBUILD installs all three.

models/LICENCE.md

@@ -134,11 +134,16 @@ declined, and the InsightFace grant of D13).
 
 | File | Source | Trained on | Used by |
 |---|---|---|---|
-| `denoise/mosaic-1408.onnx` | trained from scratch in the `darkroom-denoise` repository (2026-10-03, run `m2`, 60 000 steps) | 427 of the maintainer's own base-ISO Canon EOS 6D raws, with the 6D's measured noise added | the learned demosaic and denoise (FR-DEV-3g) |
+| `denoise/mosaic-best-1408.onnx` | trained in the `darkroom-denoise` repository (2026-10-04, run `final`, 30 000 steps, from the experts of runs `m2` and `edges-100`) | 1,701 of the maintainer's own base-ISO raws and 6,000 synthetic scenes the repository draws itself, with the Canon EOS 6D's measured noise added | the learned demosaic and denoise, Best (FR-DEV-3g) |
+| `denoise/mosaic-medium-1408.onnx` | distilled from `final` in the same repository (2026-10-04, run `student-m`, 20 000 steps, from `m2`) | the same | Medium |
+| `denoise/mosaic-fast-1408.onnx` | distilled from `final` (2026-10-04, run `student-s`, 30 000 steps, from scratch) | the same | Fast |
 
-A U-Net of plain 3×3 convolutions, ReLU, strided and transposed
+U-Nets of plain 3×3 convolutions, ReLU, strided and transposed
 convolutions and additive skips — no third-party architecture code or
-weights — at a fixed `1×1×1408×1408` for `mosaic` and `sigma`, exported by
-`python -m denoise.export` in `darkroom-denoise`. Trained only on
+weights — and, for Best, two of them blended per photosite by a small gate
+network of the same parts. Each at a fixed `1×1×1408×1408` for `mosaic`
+and `sigma`, exported by `python -m denoise.export` in `darkroom-denoise`;
+the `.a16w16.onnx` siblings are the same networks quantised for the
+Hexagon by `tools/quantise-models.sh`. Trained only on
 photographs the maintainer owns, so the weights carry no grant but the
 project's own: GPL-3.0-or-later, like the code (denoise.md §10).