Okay, there's a version 1.7.8 up now, that fetches comments using the same strategy as it fetches entries. It will grab 400 of each at a time by default, but you can set that to some other number, e.g. "--max 1500". Hopefully this will let you grab comments in batches small enough to appease the mighty server gods.

Unfortunately the database schema has changed very slightly from the earlier one, so you'll need to start the journal download from scratch again, instead of using whatever you've currently got.