warp add reduction works

juno_samples/cava/src/cava.jn

@@ -152,7 +152,7 @@ fn gamut<row : usize, col : usize, num_ctrl_pts : usize>(
         let v  = v1 * v1 + v2 * v2 + v3 * v3;
         l2_dist[cp] = sqrt!::<f32>(v);
       }
-      
+     
       @channel_loop for chan = 0 to CHAN {
         let chan_val : f32 = 0.0;
         for cp = 0 to num_ctrl_pts {

juno_samples/cava/src/gpu.sch

@@ -117,6 +117,9 @@ fixpoint {
 simpl!(fuse4);
 array-slf(fuse4);
 simpl!(fuse4);
+fork-tile[2, 0, false, true](fuse4@channel_loop);
+fork-split(fuse4@channel_loop);
+clean-monoid-reduces(fuse4);
 
 no-memset(fuse5@res1);
 no-memset(fuse5@res2);