Optimization for miranda

• Optimize cava and edge detection schedules (CPU schedules are multi-core now).
• Add fork-extend pass to make fork dimensions multiple of some number (closes #28 (closed)).
• Make GCM add extra dimensions to per-thread arrays in fork-joins.
• Make GCM allocate extra memory for calls in parallel, make RT backend use different memories per parallel call.
• Make not finding xdot binary crash the compiler.
• Some misc. fixes.
• Add conv-like fork-join test.

juno_samples/edge_detection/benches/edge_detection_bench.rs

@@ -63,6 +63,10 @@ fn edge_detection_bench(c: &mut Criterion) {
     let num_frames = 5;
 
     let mut r = runner!(edge_detection);
+    let gaussian_filter_h = gaussian_filter_h.to();
+    let structure_h = structure_h.to();
+    let sx_h = sx_h.to();
+    let sy_h = sy_h.to();
 
     let frames: Vec<_> = (0..num_frames).map(|_| load_frame(&mut video)).collect();
 
@@ -87,10 +91,10 @@ fn edge_detection_bench(c: &mut Criterion) {
                         sz as u64,
                         sb as u64,
                         input_h.to(),
-                        gaussian_filter_h.to(),
-                        structure_h.to(),
-                        sx_h.to(),
-                        sy_h.to(),
+                        gaussian_filter_h,
+                        structure_h,
+                        sx_h,
+                        sy_h,
                         theta,
                     )
                     .await