Explain why discouraged features are discouraged

[ddbeck]

The data for discouraged features ought to help developers understand why a given a feature is discouraged (or slated for removal).

Right now, discouraged features have two parts

• according_to — citations for the discouraged state (such as spec text, a change removing something from a spec, or some sort of vendor action to withdraw that feature)
• alternatives (optional) — IDs for other features which might be a substitute for some or all of the discouraged feature's substittues

Together, those might contain or imply a reason for discouragement, but some are opaque or require some insider knowledge to read between the lines. In working with MDN on presenting, they said that using web-features data for deprecation banners wouldn't work unless there was a human-readable reason.

I've filed this to give some missing context for #3193; see that PR for a possible resolution to this issue.

[ddbeck]

@atopal as mentioned in the most-recent WebDX call, this would be a good place to recap the research and user testing results for the MDN deprecated banner work.