Detect browser without relying on vendor prefixes

[jan-ivar]

Fixes #1156.

[fippo]

isn't waiting for https://bugzilla.mozilla.org/show_bug.cgi?id=1750143 easier?
That will not require a major version bump which traditionally takes years (see #1065)

[jan-ivar]

Merging this PR should have the same net effect (of making the check UA-string based).

It might also end up being faster given Mozilla's neutral position on UA client hints.

I should have done this a year ago when we discovered this, but we gotta plant the seed at some point.

[jan-ivar]

Firefox (Nightly) 138 now has media.navigator.mozgetusermedia.enabled pref in about:config that if you set it to false causes adapter to return "Not a supported browser.".

While most major video conferencing websites I've tested so far appear to work, we're in contact with one affected by this. They're applying a fix on their end, but getting this merged would probably help them and other consumers of adapter.

[fippo]

Just trying to get it into the same release as #1127

https://www.reddit.com/r/firefox/comments/1j8401w/another_media_service_fallen_f1tv_is_a_costly/?share_id=qEAEN4EyxZnU9fGVFNaDI&utm_content=2&utm_medium=android_app&utm_name=androidcss&utm_source=share&utm_term=1&rdt=56689 should have demonstrated that UA sniffing is far more costly to Firefox than helping do actual feature detection but...

[jan-ivar]

Thanks for the link. It's a good example of why we want to get rid of things like mozGetUserMedia.

Streamlining UA sniffing helps us respond quickly with out-of-band interventions or dot-releases, which might need to include overriding UA strings in cases where websites reactively blocked Firefox.

[fippo]

rebase needed

[fippo]

@jan-ivar ^

I still think that providing a minimal UserAgentData that saves developers from the pain of parsing the UA string would be easier to adopt.