fix: gamma functions

[wassup05]

This PR tries to address #898

The problem was in the interface gpx (meant for internal use only) which had two procedures for x, p are real and x is real, p is an integer, p being > 0 in all cases

Everything was fine with the second procedure but in the first one it is required that (Fast and accurate evaluation of a
generalized incomplete gamma function) when x < 0, p be an integer which wasn't actually programmed in the procedure itself but was being taken care of as and when requiring to call gpx in such a case. like so https://github.com/fortran-lang/stdlib/blob/master/src/stdlib_specialfunctions_gamma.fypp#L1078-L1099

And hence this was not found earlier as the chunk of code was in the case when x<0 and p<p_lim(x) (check Algorithm 3) which is now unreachable, hence removed

[perazz]

LGTM with the above suggestion @wassup05. Thank you for the fix.

As an additional request, it would be great to see gamma functions a little better documented in the docs files, so the lazy user doesn't have to search into the source code to understand what ranges and types they work with.

[wassup05]

Thank you @perazz for your review, I have added your suggestion in the latest commit. Do you mean docs of "in source" comment markup? Because the functions are quite well documented in the spec

[jvdp1]

LGTM. Thank you.

[jalvesz]

LGTM @wassup05

[perazz]

With 3 approvals, I will merge this PR. Thank you @wassup05 for the PR, and @jalvesz @jvdp1 for the reviews.