I find it is taking a long time to find files using the similarity
option when thousands of files are involved.
Would it make sense to be able to create the .sim files in the same way
thumbnails can be created -
e.g. I use --remote --cache-render in a cron job to make sure the
thumbnails are up-to-date
Does anyone have an idea where the run-time is being taken up - in
creating the .sim files or in comparing .sim files?
