Fix 2 CBLAS issues

[angsch]

Description

Part 1: Add the cblas interfaces related to rot that were reported as missing in #473
Part 2: Fix a linker error coming from cblas.h defining cblas_[ds]cabs1, but not having an implementation.

Checklist

• If the PR solves a specific issue, it is set to be closed on merge.

[angsch]

@martin-frbg Could you be so kind and have a look at the cblas interfaces?

I am surprised that BLAS-1 routines such as i[c,z]amax call into [d,s]cabs1 instead of following LAPACK's approach where cabs1 is reimplemented as a statement function in every relevant file. In particular for cblas_[d,s]cabs1, it feels excessive to compute |Re(.)| + |Im(.)| through 1 auxiliary subroutine and 1 function call. Is the replacement of function calls of cabs1 in favor of an inline definition something that should be addressed in the reference BLAS?

[langou]

I am surprised that BLAS-1 routines such as i[c,z]amax call into [d,s]cabs1 instead of following LAPACK's approach where cabs1 is reimplemented as a statement function in every relevant file. In particular for cblas_[d,s]cabs1, it feels excessive to compute |Re(.)| + |Im(.)| through 1 auxiliary subroutine and 1 function call. Is the replacement of function calls of cabs1 in favor of an inline definition something that should be addressed in the reference BLAS?

Good point, but that's kind of another thread. It would be good to have inline functions for these. I do not know whether there was a purposeful intent behind not using intrinsic.

[weslleyspereira]

Thanks @angsch!!

[martin-frbg]

Thanks @angsch - interfaces match what I put into OpenBLAS at the time I created the issue ticket.