Pass proper user-agent in stream_file when host is upload.wikimedia.org

[benoit74]

For files hosted on upload.wikimedia.org, we must comply with their User-Agent policy at https://meta.wikimedia.org/wiki/User-Agent_policy

Doing so at scraperlib level in stream_file (main methods using in many scraper to download files / assets) would help avoid having to do so in every scraper (and forget about it over and over).

[benoit74]

That being said, I'm not sure this is really straightforward to implement.

Scraper should pass its name and version to scraperlib so that we set properly the header

And we also need a contact, which is probably more related to who ran the scraper

Not sure this is so easy to implement in the end.