Replace BeautifulSoup with lxml or similar

[rasoolsomji]

BeautifulSoup is quite a heavyweight package to be just grabbing one JSON variable from the page. Lxml might be quicker or since understat.com is pretty well-formed you could even treat the whole document as a string and just search for the <script> tag.

I'm happy to do this and submit it as a PR if you agree it's worthwhile