[builtins] Introduce proper Float64Tan operator.

src/crankshaft/arm64/lithium-codegen-arm64.cc

Index: src/crankshaft/arm64/lithium-codegen-arm64.cc
diff --git a/src/crankshaft/arm64/lithium-codegen-arm64.cc b/src/crankshaft/arm64/lithium-codegen-arm64.cc
index ec87a1aa77d3f7bdfd729ac7abc5cff4f5b7ee2c..c4fbd5fdbeff43e8c9e79a3ced9154d9e8f931dc 100644
--- a/src/crankshaft/arm64/lithium-codegen-arm64.cc
+++ b/src/crankshaft/arm64/lithium-codegen-arm64.cc
@@ -2218,18 +2218,6 @@ void LCodeGen::DoDoubleBits(LDoubleBits* instr) {
 }
 
 
-void LCodeGen::DoConstructDouble(LConstructDouble* instr) {
-  Register hi_reg = ToRegister(instr->hi());
-  Register lo_reg = ToRegister(instr->lo());
-  DoubleRegister result_reg = ToDoubleRegister(instr->result());
-
-  // Insert the least significant 32 bits of hi_reg into the most significant
-  // 32 bits of lo_reg, and move to a floating point register.
-  __ Bfi(lo_reg, hi_reg, 32, 32);
-  __ Fmov(result_reg, lo_reg);
-}
-
-
 void LCodeGen::DoClassOfTestAndBranch(LClassOfTestAndBranch* instr) {
   Handle<String> class_name = instr->hydrogen()->class_name();
   Label* true_label = instr->TrueLabel(chunk_);