Publish plain JSON artifacts "live"

[ddbeck]

We publish a next-tagged web-features pre-release on npm on each push to main (you install it with npm install web-features@next). If you want the latest web-features data, this is convenient, unless you're not an npm user.

We ought to publish the plain JSON files on the web-features site, so that it's possible to do something like curl https://web-platform-dx.github.io/web-features/downloads/next/data.json (and likewise for the extended JSON and JSON Schema artifacts).

[tabatkins]

Yup, I was about to ask about this in #1853. My use-case is having the bikeshed-data project (which runs about 100 times/day) fetch the data, then digest it into the form that Bikeshed users will actually download. This is a Python project, and just pings a bunch of public URLs for its data sources, including currently the CanIUse and MDN data; I was hoping I could use this same strategy for Web Features data.

Note that CanIUse just has one giant JSON blob; MDN has a top-level directory listing, and then I download all the individual files it mentions. I can handle either way, whatever's convenient for y'all; all I need is a dependable way to get all the data without manual intervention.