Improved method invocation performance: 1.5x for virtual and 2.8x for interface.

- Implemented predicted chaining for invoke virtual and interface.
- Eliminated a little bit of fat for invoke native.
- Added 078-polymorphic-virtual for stress tests.
diff --git a/vm/compiler/codegen/armv5te/ArchUtility.c b/vm/compiler/codegen/armv5te/ArchUtility.c
index 7d7f119..e45c572 100644
--- a/vm/compiler/codegen/armv5te/ArchUtility.c
+++ b/vm/compiler/codegen/armv5te/ArchUtility.c
@@ -167,8 +167,11 @@
         case ARMV5TE_PSEUDO_CHAINING_CELL_HOT:
             LOGD("-------- chaining cell (hot): 0x%04x\n", dest);
             break;
-        case ARMV5TE_PSEUDO_CHAINING_CELL_INVOKE:
-            LOGD("-------- chaining cell (invoke): %s/%p\n",
+        case ARMV5TE_PSEUDO_CHAINING_CELL_INVOKE_PREDICTED:
+            LOGD("-------- chaining cell (predicted)\n");
+            break;
+        case ARMV5TE_PSEUDO_CHAINING_CELL_INVOKE_SINGLETON:
+            LOGD("-------- chaining cell (invoke singleton): %s/%p\n",
                  ((Method *)dest)->name,
                  ((Method *)dest)->insns);
             break;