Get CSS property values from webref

[queengooborg]

This PR grabs the values for CSS properties from webref, so that we can stay synchronized with new values introduced in the specification.

Statistics

Before this change, there are 1284 features in css.properties. After this change, there are 2510 features. This adds 1,226 new features and removes 0.

[teoli2003]

LGTM, let's try.

[Elchi3]

Before this change, there are 1284 features in css.properties. After this change, there are 2510 features. This adds 1,226 new features and removes 0.

This is massive. I'm not yet sure we have the bandwidth to review this. Do you know how it will affect our coverage (which is a goal)?

Also, what is the plan for avoiding to introduce a lot of ranged values? For example if you have the CSS height property supported since version 1 and you add a sub feature for the height: auto value, which I think was a value that was also added in version 1, how will the collector do this? I often see the collector setting ≤72 for Firefox for example as the earliest version. This is problematic because MDN renders it as "72" which is not ideal as I think we should be indicating that height and height: auto were introduced at the same time, in version 1.

[queengooborg]

Do you know how it will affect our coverage (which is a goal)?

A natural side effect of adding these features is that our coverage technically increases, because we're adding more covered features to BCD. Before v10.7.0 (specifically, commit 1a08c73), the coverage for css.properties was 213/1227 (17.36%). After v10.7.0 was released (plus a few tweaks, 1adc979), then running add-new-bcd, coverage went down to 206/1784 (11.55%).

Now, if we're focusing only on the coverage improvement of features already tracked in BCD, well, it's only 7 features (a 0.57% difference) -- not exactly a grand improvement. The intention of this change isn't the coverage, but rather improving BCD by adding verbosity about what values we know are supported. If we're just focused on the numbers, then it's a plus for coverage.

---

Also, what is the plan for avoiding to introduce a lot of ranged values?

I didn't really have a plan for that since we have still been focusing on only correcting errors for browser versions released in 2020 and later. However, it's an easy fix: all we need to do is collect results for older browsers, which is as simple as changing a command line argument.

[Elchi3]

Thanks for your analysis! I will not review new CSS property values data additions for the moment then. Need to focus on the other goals. I think this would be a great project later in the year, though.