packager should tell the user in its CLI output that a reference to Article is missing from METADATA.pb

Background: Related to https://github.com/google/fonts/pull/7875, if the METADATA.pb file in a PR doesn't have a `files {}` stanza pointing to the `ARTICLE.en_us.html` source file, the packager command creates an empty one. The empty one then triggers a FAIL in Fontbakery that is obtuse for the user, because it complains about a missing HTML tag in the ARTICLE.en_us.html file (and the HTML tag in the Fontbakery report is un-escaped, so it's invisible in the message that it posts to the GitHub report).

The underlying problem there is that the file isn't referenced in METADATA.pb; the packager could report that problem to the user on the command line, like it does for other serious warnings before it generates the PR. Then the user could fix the central issue immediately.

This (line four) is the message printed by packager when it generates the empty ARTICLE file because it doesn't see it in METADATA.pb:


That "Wrote" line does not look like an error. Especially because the ARTICLE.en_us.html file _is_ in the repo. Compare to:

It might just require tweaking the output. The message printed would be more helpful if it was aligned to the others, with a timestamp and so on (because it kind of looks like it's line-wrapping overflow, rather than a message), and IMO it should use highlighting a la "Warning". But if the empty file is going to cause Fontbakery to fail, I think it should provide a clearer message, telling the user that it was not finding the `file {}` entry in the METADATA.pb file, rather than just reporting the presence/creation of the file -- because editing the entry in METADATA.pb is the fix it needs.

(For completion, the message printed when the ARTICLE file _is_ found via the METADATA.pb `file{}` is also not lined up with the other CLI stdout output; changing that too might also help a tad, since users would grow accustomed to seeing it:

)