XPaths: improve accuracy for major news outlets

[adbar]

This PR improves the accuracy of main content detection (especially before and after the actual content) for the following websites:

• AP News
• Free Beacon
• LA Times
• Reuters
• Washington Times

Credit to fundus-evaluation for evaluating the software and providing a benchmark.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.35%. Comparing base (fa972ab) to head (fdf1f2f).

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #530   +/-   ##
=======================================
  Coverage   97.35%   97.35%           
=======================================
  Files          22       22           
  Lines        3439     3439           
=======================================
  Hits         3348     3348           
  Misses         91       91           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[adbar]

@felipehertzer I'm leaving this PR open for now, if you want to have a look or add something feel free to do so.

[felipehertzer]

Hey @adbar, I will have a look this week, the changes makes sense.

But, I have two websites from the last pull request that I couldn't find a way to fix it.

Both are getting more text that should be. If you want to have a look:
Website 1

Website 2

[adbar]

@felipehertzer The first extracts too much content indeed but the second works nearly OK.

I'll merge the PR now as I'll release a new version shortly but you can have a look at it later and suggest improvements if you find something. The benchmark above is also interesting.