Implementing client hints support

[DamonOehlman]

A big thanks to @scottcanoni who fired through an email to me flagging the upcoming deprecation and freezing of the Chrome UA string, which will mean that detection of Chrome will be version locked at Chrome 85 as we move into the future.

see: https://groups.google.com/a/chromium.org/forum/#!msg/blink-dev/-2JIRNMWJ7s/yHe4tQNLCgAJ

Now I've had a quick gloss over the client hints API (what's coming into replace user agent strings) and I think what is being done there looks excellent. More details on what is being exposed to the browser can be found at the following URL:

https://wicg.github.io/ua-client-hints/#create-brands

So what's the impact of this on detect-browser?

In the short term, this package should help to ease the transition as client hints are rolled out across browsers, i.e. it should continue to work as it does, returning the same structured responses but using client hints to get that information.

Longer term, however, it should probably be deprecated. This is, of course, assuming that browser vendors do a good job of the implementation and developers can rely on those client hints for accurate information. If in the event that this doesn't happen, OR, we consider the value of having a discrete set of known values for browsers (provided by the typescript literal union) valuable detect-browser will continue to exists. This is a larger discussion and out of scope for the immediate work, but worth raising now. My personal view is that a package should only live as long as it is needed, and then point people towards the appropriate (and often better) solution that has been implemented.