Additional compilers

[hiker]

This just adds declarations for additional compilers:

• intel llfcm (icx, ifx)
• nvidia (nvc, nvfortran)
• cray (craycc, crayftn)

And also supports the Cray compiler wrapper: on a Cray system, the compilers are always called cc/ftn - but these are just wrappers that (depending on the loaded modules) will execute either the corresponding gnu, intel, or gnu tool (with additional parameters to automatically link libraries like MPI based on the modules loaded.

Since I didn't want to define Ftn and Cc as compiler wrapper (while ftn is pretty much cray specific, cc is definitely very common and would result in confusion), I have therefore names the compiler wrapper CrayFtn and CrayCc (which will be used to create the compiler wrapper for intel and gnu, e.g. with the names crayftn-ifort, or craycc-gcc).

The actual 'native' compiler are only called using the wrapper script, so in order to have a consistent naming scheme, I used the corresponding names crayftn-ftn and craycc-cc - and the classes are names Crayftn and Craycc. That's a bit confusing (given that CrayFtn etc are wrapper), but it was the best idea I had to use class names and compiler (wrapper) names consistently.

Better ideas are welcome :)

[hiker]

I also removed the whole concept of mixins to handle the compiler version detection. I am aware that I originally suggested it :) But it turns out that in all these cases so far a simple regex is all that is needed, and it additionally simplifies the code (no more need in the mixins to check which compiler type there is to detect the compiler name)