If you select all files in the database and run the dupe search, the result window shows all originals with their duplicates.
Files without duplicates are not listed.
IMatch cannot decide which of the files you want to keep. If you switch the result window to show the "original" files in-line with their results, you can easily view/delete/bookmark/copy/cut ... any file, like in other file windows.
In most cases you can use the file window search bar or the filter panel or one of the workflow categories (like, "no keywords" or "no title") to filter out the unwanted and so-far unprocessed duplicates. This depends on your setup of course, and why you have duplicates, how the duplicates differ from files you want to keep. I'm sure you can work something out. 32K files is not much. I did this with collections of 500,000 files several times.
If you don't care and just want to keep one file, don't in-line the originals with matches. In the result window then just do a Ctrl+A and <Del> to reject all dupes.
Review the files in the "Rejected" collection and then delete.