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/fork_join_tests/src/cpu.sch

@@ -3,7 +3,7 @@ gvn(*);
 phi-elim(*);
 dce(*);
 
-let auto = auto-outline(test1, test2, test3, test4, test5, test7, test8);
+let auto = auto-outline(test1, test2, test3, test4, test5, test7, test8, test9);
 cpu(auto.test1);
 cpu(auto.test2);
 cpu(auto.test3);
@@ -11,6 +11,7 @@ cpu(auto.test4);
 cpu(auto.test5);
 cpu(auto.test7);
 cpu(auto.test8);
+cpu(auto.test9);
 
 let test1_cpu = auto.test1;
 rename["test1_cpu"](test1_cpu);
@@ -51,11 +52,11 @@ fixpoint panic after 20 {
   unroll(auto.test1);
 }
 
-fork-split(auto.test2, auto.test3, auto.test4, auto.test5);
+fork-split(auto.test2, auto.test3, auto.test4, auto.test5, auto.test9);
 gvn(*);
 phi-elim(*);
 dce(*);
-unforkify(auto.test2, auto.test3, auto.test4, auto.test5);
+unforkify(auto.test2, auto.test3, auto.test4, auto.test5, auto.test9);
 ccp(*);
 gvn(*);
 phi-elim(*);
@@ -93,4 +94,6 @@ dce(auto.test8);
 simplify-cfg(auto.test8);
 dce(auto.test8);
 
+no-memset(test9@const);
+
 gcm(*);