NEMO concept is to convert homegeneous sets of files (same information at the same position in all the files of one set).
So the number of file is not a problem for NEMO, the problem is there heterogeneity.
If you are able to divide your huge set of files into a small (10?, 20?, 30?) number of groups of homogeneous files,then you'll have to define models only for each groups of files and run NEMO in batch mode.
If you are not able to make homogeneous groups of files, then using NEMO is possible but much more time consuming, because you'll have to define a model per file.