Examine why repology has different results for MetaCPAN vs CPAN

[oalders]

See repology/repology-webapp#217 for context. Once this had been appropriately addressed, we can see about having repology drop CPAN in favour of having just MetaCPAN.

[Leont]

Both being shadow repositories makes it a bit more difficult (it means that packages aren't shown unless they're also packaged by a repository that isn't shadowed). That can make the data a bit confusing.

I have the impression that one differences is that MetaCPAN is a backpan, and CPAN is not.

[AMDmi3]

Both being shadow repositories makes it a bit more difficult

Would you like me to unshadow them, or prepare a dedicated dataset?