I'm a software development engineer. If this stuff is media data that could entail many formats - I would say the least likely would be csv imo. If it were from a database, I don't think it would be beyond the bounds of reason that the FBI are capable of taking a database dump and loading it into their own DB instance whether it be Oracle, MySql, SQL Server etc, I'd say it was unlikely it would be asked to be transferred in csv format - that would be crazy, not impossible but why do it that way when you can get a copy of their whole database (I guess their could be some issues regarding the database schema design itself being proprietary - so only allowed to take the data, not the database?).
Anyway, even csv files are amenable to plain text keyword searches. I would think the team would be looking for some specific things. If there were csv files from a database they would be per Db table, it wouldn't take long to work out the basic DB schema, or even load them into a spreadsheet. Searching for specific Russian words in html, xml, csv, asp, txt, xsl, word, even from a large dump of files is not difficult or too time consuming to get a handle into the data.
Other file types that are binary would require software to read them, unless they are encrypted or zipped where presumably they would return to the source to get them into a readable format. But the file type can easily be determined and the relevant application found to read them in two ticks.
IMO you wouldn't even have to translate all the files that contain textual information. You search for the Russian keywords you are interested in and translate the hits.
I think this whole question about dumping large amounts of data is a distraction. Presumably if going through all that information is a difficult time consuming job, it's the same problem for both prosecution and defense anyway.