[builtins] Introduce proper Float64Atan and Float64Atan2 operators.

src/compiler/machine-operator-reducer.cc

Index: src/compiler/machine-operator-reducer.cc
diff --git a/src/compiler/machine-operator-reducer.cc b/src/compiler/machine-operator-reducer.cc
index c4d0aec31c042513a3278b6ee9db985f9a047cbf..dddf14d0b081afec940f90ee9574f8b6dbfe4712 100644
--- a/src/compiler/machine-operator-reducer.cc
+++ b/src/compiler/machine-operator-reducer.cc
@@ -383,6 +383,25 @@ Reduction MachineOperatorReducer::Reduce(Node* node) {
       }
       break;
     }
+    case IrOpcode::kFloat64Atan: {
+      Float64Matcher m(node->InputAt(0));
+      if (m.HasValue()) return ReplaceFloat64(base::ieee754::atan(m.Value()));
+      break;
+    }
+    case IrOpcode::kFloat64Atan2: {
+      Float64BinopMatcher m(node);
+      if (m.right().IsNaN()) {
+        return Replace(m.right().node());
+      }
+      if (m.left().IsNaN()) {
+        return Replace(m.left().node());
+      }
+      if (m.IsFoldable()) {
+        return ReplaceFloat64(
+            base::ieee754::atan2(m.left().Value(), m.right().Value()));
+      }
+      break;
+    }
     case IrOpcode::kFloat64Log: {
       Float64Matcher m(node->InputAt(0));
       if (m.HasValue()) return ReplaceFloat64(base::ieee754::log(m.Value()));