ART: Generate chained compare-and-branch for short switches

Refactor Mir2Lir to generate chained compare-and-branch sequences
for short switches on all architectures.

Bug: 16241558

(cherry picked from commit 48971b3242e5126bcd800cc9c68df64596b43d13)

Change-Id: I0bb3071b8676523e90e0258e9b0e3fd69c1237f4
diff --git a/compiler/dex/quick/arm/codegen_arm.h b/compiler/dex/quick/arm/codegen_arm.h
index 9078298..cd6c9cc 100644
--- a/compiler/dex/quick/arm/codegen_arm.h
+++ b/compiler/dex/quick/arm/codegen_arm.h
@@ -131,8 +131,8 @@
                                        int first_bit, int second_bit);
     void GenNegDouble(RegLocation rl_dest, RegLocation rl_src);
     void GenNegFloat(RegLocation rl_dest, RegLocation rl_src);
-    void GenPackedSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src);
-    void GenSparseSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src);
+    void GenLargePackedSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src);
+    void GenLargeSparseSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src);
 
     // Required for target - single operation generators.
     LIR* OpUnconditionalBranch(LIR* target);