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

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

Change-Id: Ie2a572ae69d462ba68a119e9fb93ae538cddd08f
diff --git a/compiler/dex/quick/x86/codegen_x86.h b/compiler/dex/quick/x86/codegen_x86.h
index 266191a..d74caae 100644
--- a/compiler/dex/quick/x86/codegen_x86.h
+++ b/compiler/dex/quick/x86/codegen_x86.h
@@ -246,8 +246,8 @@
                                      int first_bit, int second_bit) OVERRIDE;
   void GenNegDouble(RegLocation rl_dest, RegLocation rl_src) OVERRIDE;
   void GenNegFloat(RegLocation rl_dest, RegLocation rl_src) OVERRIDE;
-  void GenPackedSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src) OVERRIDE;
-  void GenSparseSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src) OVERRIDE;
+  void GenLargePackedSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src) OVERRIDE;
+  void GenLargeSparseSwitch(MIR* mir, DexOffset table_offset, RegLocation rl_src) OVERRIDE;
 
   /**
    * @brief Implement instanceof a final class with x86 specific code.