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/gpu.sch

@@ -7,12 +7,13 @@ no-memset(test3@const3);
 no-memset(test6@const);
 no-memset(test8@const1);
 no-memset(test8@const2);
+no-memset(test9@const);
 
 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);
 gpu(auto.test1);
 gpu(auto.test2);
 gpu(auto.test3);
@@ -20,6 +21,7 @@ gpu(auto.test4);
 gpu(auto.test5);
 gpu(auto.test7);
 gpu(auto.test8);
+gpu(auto.test9);
 
 ip-sroa(*);
 sroa(*);
@@ -34,7 +36,7 @@ fixpoint panic after 20 {
 }
 
 fixpoint panic after 20 {
-  fork-coalesce(auto.test1, auto.test3, auto.test4, auto.test5, auto.test7, auto.test8);
+  fork-coalesce(auto.test1, auto.test3, auto.test4, auto.test5, auto.test7, auto.test8, auto.test9);
 }
 
 gvn(*);
@@ -71,6 +73,8 @@ dce(auto.test8);
 simplify-cfg(auto.test8);
 dce(auto.test8);
 
+no-memset(test9@const);
+
 ip-sroa(*);
 sroa(*);
 dce(*);
@@ -78,7 +82,7 @@ ccp(*);
 gvn(*);
 phi-elim(*);
 dce(*);
-gcm(*);
 
+gcm(*);
 float-collections(test2, auto.test2, test4, auto.test4, test5, auto.test5);
 gcm(*);