Just read your post. I did a near identical setup as yours for the large data migration (over 5 million documents, from multiple Lotus Notes databases & forms) about 5 years ago, from Lotus Notes to IBM DB2.
As part of that migration we also migrated & sync'd data records from a number of MS Excel spreadsheets, MS access databases, Lotus Approach databases, Lotus 123 spreadsheets and MS Word tables.
Had the same issues with the RichText migration. We chose to automatically print all the Notes documents to PDF files (to retain the format) and then just migrated the unformatted text.
We had many issues, as we were migrating multiple Lotus Notes databases (built by multiple dev's, over numerous years) and other data sources, which required a lot of data cleansing reports for missing fields, incorrect data types, etc...
Also, given that we moving to a new system (IBM Websphere and DB2), there was a lot of work required in data mapping between the old and new systems. We had to provide mechanisms for the SME's to do this mapping, with the results being used to for the final migration to the new system.
Feel free to give me a shout some time and we can compare notes. I may have a few things, tips, idea's; which may help you.