fix(cnn_v3): remove dec0 ReLU, load FiLM MLP at runtime

Two bugs blocking training convergence:

1. dec0 ReLU before sigmoid constrained output to [0.5,1.0] — network
   could never produce dark pixels. Removed F.relu in train_cnn_v3.py
   and max(0,…) in cnn_v3_dec0.wgsl. Test vectors regenerated.

2. set_film_params() used hardcoded heuristics instead of the trained MLP.
   Added CNNv3FilmMlp struct + load_film_mlp() to cnn_v3_effect.h/.cc.
   MLP auto-loaded from ASSET_WEIGHTS_CNN_V3_FILM_MLP at construction;
   Linear(5→16)→ReLU→Linear(16→72) runs CPU-side each frame.

36/36 tests pass. Parity max_err=4.88e-4 unchanged.

handoff(Gemini): retrain from scratch — needs ≥50 samples (currently 11).
See cnn_v3/docs/HOWTO.md §2-3.

1 files changed, 1 insertions, 1 deletions

diff --git a/cnn_v3/shaders/cnn_v3_dec0.wgsl b/cnn_v3/shaders/cnn_v3_dec0.wgsl
index 617b5a2..79fd837 100644
--- a/cnn_v3/shaders/cnn_v3_dec0.wgsl
+++ b/cnn_v3/shaders/cnn_v3_dec0.wgsl
@@ -64,7 +64,7 @@ fn dec0_main(@builtin(global_invocation_id) id: vec3u) {
                 }
             }
         }
-        let v = max(0.0, params.gamma[o] * sum + params.beta[o]);
+        let v = params.gamma[o] * sum + params.beta[o];
         out[o] = 1.0 / (1.0 + exp(-v));
     }