This PR adds functionality needed for QUDA to Jitify. Once merged into the v1 branch, it should propagated to v2 since this is also a critical PR for RAPIDS which uses the v2 branch.

• Extends the functionality of type_traits with
  • is_floating_point_v (c 17)
  • is_integral_v (c 17)
  • is_same_v (c 17)
  • is_pointer, is_pointer_v (c 17)
  • remove_pointer
  • is_arithmetic and is_arithmetic_v (c 17)
• make all now builds jitify_test as well as jitify_example
• Fix: the presence of __LP64__ should be checked for when deciding on word size
• Fix: align the definitions of the climits built in to match those in libcu
• Add test BuiltinNumericCudaStdLimitsHeader which mimics BuiltinNumericLimitsHeader but uses libcu