Update fetch API to improve reliability

[aganders3]

This draft includes two proposed changes:

1. Fetch available plugins via the PyPI XML-RPC API, which is generally much faster and seems more reliable than scraping, but is considered legacy and planned to be deprecated.
2. Fallback on scraping as we do now, but with improved reliability by explicitly setting the ordering to "Date last updated" which results in more stable results.

See #262 for more discussion on whether the first is desirable - I will leave this as draft until that issue is resolved, but I think the second change should be incorporated either way.

[aganders3]

N.B. this is already a transitive dependency I think

[codecov]

Codecov Report

Merging #265 (12c8093) into main (1c51f58) will decrease coverage by 0.07%.
The diff coverage is 92.30%.

@@             Coverage Diff             @@
##              main     #265      +/-   ##
===========================================
- Coverage   100.00%   99.92%   -0.08%     
===========================================
  Files           37       37              
  Lines         2814     2833      +19     
===========================================
+ Hits          2814     2831      +17     
- Misses           0        2       +2     

Impacted Files	Coverage Δ
npe2/_inspection/_fetch.py	99.08% <92.30%> (-0.92%)	⬇️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.