[GlobalISel] Add IRTranslator support for G_FNEARBYINT

[paquette]

Summary

Translate llvm.nearbyint into G_FNEARBYINT as a simple intrinsic. Update arm64-irtranslator.ll.

[paquette]

Pushing this along because it's trivial and Amara isn't around right now

llvm/lib/CodeGen/GlobalISel/IRTranslator.cpp

switch (ID) {
     case Intrinsic::fma:
       return TargetOpcode::G_FMA;
     case Intrinsic::log:
       return TargetOpcode::G_FLOG;
     case Intrinsic::log2:
       return TargetOpcode::G_FLOG2;
     case Intrinsic::log10:
       return TargetOpcode::G_FLOG10;
+    case Intrinsic::nearbyint:
+      return TargetOpcode::G_FNEARBYINT;
     case Intrinsic::pow:
       return TargetOpcode::G_FPOW;
     case Intrinsic::round:
       return TargetOpcode::G_INTRINSIC_ROUND;
     case Intrinsic::sin:
       return TargetOpcode::G_FSIN;
     case Intrinsic::sqrt:
       return TargetOpcode::G_FSQRT;

llvm/test/CodeGen/AArch64/GlobalISel/arm64-irtranslator.ll

 declare float @llvm.floor.f32(float)
 define float @test_floor_f32(float %x) {
   ; CHECK-LABEL: name:            test_floor_f32
   ; CHECK: %{{[0-9]+}}:_(s32) = G_FFLOOR %{{[0-9]+}}
   %y = call float @llvm.floor.f32(float %x)
   ret float %y
 }
 
+declare float @llvm.nearbyint.f32(float)
+define float @test_nearbyint_f32(float %x) {
+  ; CHECK-LABEL: name:            test_nearbyint_f32
+  ; CHECK: %{{[0-9]+}}:_(s32) = G_FNEARBYINT %{{[0-9]+}}
+  %y = call float @llvm.nearbyint.f32(float %x)
+  ret float %y
+}
+
 ; CHECK-LABEL: name: test_llvm.aarch64.neon.ld3.v4i32.p0i32
 ; CHECK: %1:_(<4 x s32>), %2:_(<4 x s32>), %3:_(<4 x s32>) = G_INTRINSIC_W_SIDE_EFFECTS intrinsic(@llvm.aarch64.neon.ld3), %0(p0) :: (load 48 from %ir.ptr, align 64)
 define void @test_llvm.aarch64.neon.ld3.v4i32.p0i32(i32* %ptr) {
   %arst = call { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0i32(i32* %ptr)
   ret void
 }
 
 declare { <4 x i32>, <4 x i32>, <4 x i32> } @llvm.aarch64.neon.ld3.v4i32.p0i32(i32*) #3