Ranking is not enough centered on Perl

[dolmen]

The "About" page says: « "Github Meets CPAN" presents a list of users who have set their Github account at MetaCPAN. »
So I expect that it ranks CPAN authors accordingly to some statistics related to Perl development on GitHub.

However it doesn't seem to be the case. General Github statistics are used instead of Perl related ones.
For example, James Turnbull is ranked #17 while he has only 1 CPAN module and 0 Perl projects on GitHub.

Here are some suggestion for tweaking the ranking formula:

• count only Perl projects (instead of all projects): Github repos for which the main language is Perl
  • watchers
  • forks
• count only followers that are interested in Perl: either have a CPAN ID or have themselves at least one repo whose main language is Perl
• do not count watchers who are also followers

[plu]

Makes sense, I have merged your pull request, thanks :).

[y]

This update only looks at the language that GitHub thinks is the primary language, but the Pithub API shows that there is a way to examine all languages for a repo. A few examples where this matters: GitHub frequently misidentifies XS modules as C; and if there's JavaScript included in a repo, GitHub can identify the primary language as JavaScript if it has more lines of code. To avoid this issue, all the languages for a repo should be examined to see if Perl occurs in the list and not just as the primary language.

[oalders]

Closing this as the outstanding issue to address is captured in #4