Using user-supplied file extension and/or mime type

Usually, I don't have a completely opaque array of bytes I need to analyze. I have some kind of user-supplied filename or extension. It would be nice if magika could make use of them, especially for lower confidence results.

The issue names no files, tests, or entry points, so first trace how Magika accepts input and reports confidence. Define how a supplied filename extension or MIME type should influence low-confidence results, including precedence and fallback behavior, then add tests for the agreed cases.

Written by the indexing model from the issue text.