I've tested a fix on the develop branch that reports an error where a mimeinfo signature file duplicates IDs. In doing so, discovered both the freedesktop.org.xml and tika-mimetypes.xml files have duplicate IDs.
In both cases, these appear to be errors in those files:
in the freedesktop case, the second entry for font/woff should be font/woff2
in tika's case, the whole entry for application/x-gtar is duplicated for some reason.