[builtins] Introduce proper Float64Tan operator.

src/crankshaft/x64/lithium-x64.h

Index: src/crankshaft/x64/lithium-x64.h
diff --git a/src/crankshaft/x64/lithium-x64.h b/src/crankshaft/x64/lithium-x64.h
index 5388a7b24497be9a3ee3c10b95e119e0900f97ad..64071ccee0621e770077b3f4fd2f280e4c584bd6 100644
--- a/src/crankshaft/x64/lithium-x64.h
+++ b/src/crankshaft/x64/lithium-x64.h
@@ -53,7 +53,6 @@ class LCodeGen;
   V(ConstantI)                               \
   V(ConstantS)                               \
   V(ConstantT)                               \
-  V(ConstructDouble)                         \
   V(Context)                                 \
   V(DebugBreak)                              \
   V(DeclareGlobals)                          \
@@ -2350,20 +2349,6 @@ class LDoubleBits final : public LTemplateInstruction<1, 1, 0> {
 };
 
 
-class LConstructDouble final : public LTemplateInstruction<1, 2, 0> {
- public:
-  LConstructDouble(LOperand* hi, LOperand* lo) {
-    inputs_[0] = hi;
-    inputs_[1] = lo;
-  }
-
-  LOperand* hi() { return inputs_[0]; }
-  LOperand* lo() { return inputs_[1]; }
-
-  DECLARE_CONCRETE_INSTRUCTION(ConstructDouble, "construct-double")
-};
-
-
 class LAllocate final : public LTemplateInstruction<1, 2, 1> {
  public:
   LAllocate(LOperand* context, LOperand* size, LOperand* temp) {