SD Card imports are very slow. much slower than what reading from the SD card should be.
I suspect that the importer checks each file using either or more of
a) file size
to determine which files to import and which are duplicates,
Ny suspicion is that the checksumming reads the while file.
That sounds counterproductive. It should really just checksum an initial 1K or such
and only if that matches that of another candidate file, go one order of magnitude larger (1k, then 100k, then 1B etc) until it finds a difference or hass checksummed the whole file
That would greatly speed up imports I gather
(or course it would need to keep a database of the relevant checksums with each imported file)