Update ascii - second round: clean up the code and implement robust versions of to_lower and to_upper

[arjenmarkus]

The comments to the previous pull request are now all solved:

• Extra parentheses have been removed (for readability those around the logical disjunctions and conjunctions have been kept)

• Some clean-up of the comments (more explicitly stating the ranges)

• The functions to_lower and to_upper have been made robust (perhaps slower than before, but the performance is certainly not bad)

Suggestion for further improvement: allow to_lower and to_upper to convert strings of arbitrary length - this seems to me to be the most common usage.

[ivan-pi]

Suggestion for further improvement: allow to_lower and to_upper to convert strings of arbitrary length - this seems to me to be the most common usage.

This has been requested as part of the strings module. I have wondered already if it would be easier to just have a single function with the interface:

pure function upper(str) result(ustr)
character(len=*), intent(in) :: str
character(len=len(str)) :: ustr
end function

and get rid of the single character case conversions entirely. As you say, the most common usage will be arbitrary length strings, so I am not sure if we need a specialization for single letters. The arbitrary length version can be used on single characters too (hopefully the compilers are capable enough to remove the loop for characters with len=1).

The reason I included the case conversions in the first plase was just to cover the same functionality as the <ctype.h> header file from C.

[milancurcic]

@arjenmarkus @ivan-pi @14NGiestas Is this PR good to go or should this comment still be addressed?

[arjenmarkus]

Argh, I had intended to rely to these remarks, but never got around to it. I agree: in C the situation is completely different - there are characters and there are arrays of characters. In Fortran we have character strings but not a separate character type. The suggested interface (and almost trivial implementation) looks good to me.

[milancurcic]

Okay, I will merge this then. The suggested interface can go in as a new PR.