clang/HIP: Use abs builtins instead of implementing them

[arsenm]

Summary

InstCombine already put these back together so there's no visible
change in the -O1 test for the header.

[AlexVlx]

LGTM, thanks.

[arsenm]

43f314f5e6cebe02ff63d5197c8e5c25204b20d2

clang/lib/Headers/__clang_hip_math.h

 float __tanf(float __x) {
   return __sinf(__x) * __builtin_amdgcn_rcpf(__cosf(__x));
 }
 // END INTRINSICS
 
 #if defined(__cplusplus)
 __DEVICE__
 int abs(int __x) {
-  int __sgn = __x >> (sizeof(int) * CHAR_BIT - 1);
-  return (__x ^ __sgn) - __sgn;
+  return __builtin_abs(__x);
 }
 __DEVICE__
 long labs(long __x) {
-  long __sgn = __x >> (sizeof(long) * CHAR_BIT - 1);
-  return (__x ^ __sgn) - __sgn;
+  return __builtin_labs(__x);
 }
 __DEVICE__
 long long llabs(long long __x) {
-  long long __sgn = __x >> (sizeof(long long) * CHAR_BIT - 1);
-  return (__x ^ __sgn) - __sgn;
+  return __builtin_llabs(__x);
 }
 #endif
 
 __DEVICE__
 float acosf(float __x) { return __ocml_acos_f32(__x); }
 
 __DEVICE__
 float acoshf(float __x) { return __ocml_acosh_f32(__x); }