More math functions

[dalg24]

Fix #4765
Partially resolving #4767

Adding round, rint, logb, nextafter, copysign, and signbit
Adding round, lround*, llround**, rint, lrint*, llrint**, logb, nextafter, copysign, and signbit
Naming functions that are missing in comments
** are missing with the SYCL backend

[masterleinad]

Looks OK to me. I checked that the respective SYCL functions are indeed not implemented.
We should add unit tests, though.

[dalg24]

Withdrawing the functions that are prefix with l and ll and that are not provided by SYCL. I overlooked that their return type was long and long long respectively. Will push tests shortly.

[dalg24]

signbit is not tested (not trying to sweep it under the carpet but would rather withdraw than write all the boilerplate code needed)

[dalg24]

Failure with Clang+CUDA

4: rint(int)
4: rint(long)
4: rint(long long)
4: rint(unsigned int)
4: rint(unsigned long)
4: rint(unsigned long long)
4: rint(float)
4: relative difference exceeds tolerance [5.000000e-01 > 1.192093e-07]
4: relative difference exceeds tolerance [5.000000e-01 > 1.192093e-07]
4: value at 2.500000 which is 3.000000 was expected to be 2.000000
4: value at -2.500000 which is -3.000000 was expected to be -2.000000
4: /var/jenkins/workspace/Kokkos/core/unit_test/TestMathematicalFunctions.hpp:447: Failure
4: Expected equality of these values:
4:   errors
4:     Which is: 2
4:   0
4: rint(double)
4: relative difference exceeds tolerance [5.000000e-01 > 2.220446e-16]
4: relative difference exceeds tolerance [5.000000e-01 > 2.220446e-16]
4: value at 2.500000 which is 3.000000 was expected to be 2.000000
4: value at -2.500000 which is -3.000000 was expected to be -2.000000
4: /var/jenkins/workspace/Kokkos/core/unit_test/TestMathematicalFunctions.hpp:447: Failure
4: Expected equality of these values:
4:   errors
4:     Which is: 2
4:   0
4: [  FAILED  ] cuda.mathematical_functions_nearest_interger_floating_point_operations (3 ms)
4: [ RUN      ] cuda.mathematical_functions_floating_point_manipulation_functions

[dalg24]

On 2nd thought withdrawing rint for now since CUDA does not honor the dynamically-set rounding mode. We can revisit later.

[dalg24]

Retest this please

[masterleinad]

Looks OK to me,

[dalg24]

Failure is clearly unrelated (clang crashing when compiling TestOpenMPTarget_Complex.cpp)