Remove lxml in favor of bs4

[jsangmeister]

The breaking changes of lxml version 5.2 (seen in #2351) pointed me to the fact that we are currently using two tools (lxml and bs4) for the same purpose (extracting plain text from html content). We should conglomerate these and not use two tools without a good reason (which I don't see in this case). Since bs4 seems to be the more versatile, forgiving and easier-to-use parser, I would opt to remove lxml.