Yes. The reason to have file entities is mainly to have a platform that can store semantic descriptions of a file. For text searches in classical terms it doesn't matter much, but to search things like:
- "portrait engravings by artists born in Dordrecht"
- "depictions of Dutch poets born between 1600 and 1700"

For these kind of searches, the only possible way to return relevant results is to store the information a semantic way as Wikidata does.
As Thomas pointed out, the task to transition to the new method looks somewhat daunting, luckily here there is not much trouble using bots to automate the task filling out the properties of the 17M files.
The case of "image promotion" I think it is a different issue that would require some tagging (maybe "best depiction of") or a simple voting system (like in youtube, reddit, etc).

It is also important to note that the old issue of sexual content in Commons [1] has gained *a lot* of traction lately since the last three op-ed's questioning/defending its suistainability [2] [3]. Basically there is a need that the searches show what you are looking for and not some other random content. The urgency to present a solution is very high at the moment, a matter of weeks before starting organizing WikiLoveMonuments with a cleaned reputation, so I hope that Wikidata can present a proposal soon that I am sure will be better than this other proposal [4]

Cheers,
Micru

[1] https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2010-05-10/Commons_deletions
[2] https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2013-06-12/Op-ed
[3] https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2013-06-19/Op-ed
[4] https://www.mediawiki.org/wiki/Requests_for_comment/Image_information