Improve stripping of source from parsed headline

[r1b]

The problem:

Many headlines (esp. from title tags) have a prelude / postlude with a delimeter and the name of the site e.g:

Man Dies in Bathtub - Associated Press
Reuters | Dormant Volcano Found in Bavarian Woman's Backyard

I have a heuristic that strips these out but it is unreliable. Further, sometimes you actually want the info after the delimeter, e.g:

"The U.S.A has a fake news problem" - Trump

[r1b]

For now I am going to remove the code that does this. The only thing I will do to the headline is remove leading / trailing whitespace.

[r1b]

I will also add headline selectors for all sources - should help mitigate the issue.