When viewing HTML contents under the /content/raw/ endpoint of the web app, we currently return text/plain as MIME type in the HTTP response. See, e.g.:

$ curl -s --head https://archive.softwareheritage.org/browse/content/sha256:1d727fc911579e5ec2d6cd463189cdb76bf7ef6c8f487f1a0701a1ec06aabdf1/raw/ | grep -i Content-Type
Content-Type: text/plain

This means we currently have no way of allowing users to render HTML pages in-browser, short of downloading them and open them locally --- which is not very practical.

Rendering the HTML within the webapp is not terribly safe, so that's probably out of question.

On the other hand, returning a best-guess Content-Type when browsing content via /content/raw might be an option?
And probably we can just reuse the libmagic-detected MIME type that we already use for other purposes?

(thanks to Konrad Hinsen for this feature request)