This is the KB I was going to find today and post. I think it is worth noting that these types pre-date the newest indexing technologies.
For me, it is more about indexing than storing. If the file is in a supported path and reasonable (like not 100GB file... although I have seen craziness there too) it should probably pick it up regardless of size. Indexing is the challenge.
The KB provided is likely still mostly applicable IMHO because it consists mostly of files that do not make sense to index... or that require additional product capabilities. I seem to recall some native OCR capabilities being introduced, and that may be after the period of time this KB was posted.
There were odd quirks historically, like if you archived EML or MSGs off a File Server, they represented as if they were from Exchange in EV, but they are pretty edge cases, and most of them were sorted per my recollection. Enhanced software means new opportunities for stuff to be busted.
As the KB is published, it should still be valid or challenged if not.
