I found a work-around for this problem, if anyone else comes across it. If you do have some MailSteward records that do not import directly into EagleFiler, the best way to get them in is to export the problematic records from MailSteward to the mbox format, and then import them into Mail.app. Now you might notice that you have double the count that you should. This is due to MailSteward’s fault. Fortunately, the duplicates will all be blank, and as far as I can tell, the non-blank ones are complete emails with no problems in them. The headers are formed correctly with the original dates, and attachments are present.
Since the duplicates are blank, you can easily sort them all by clicking on the From column in Mail. This column should never be blank, so it is safer than searching by Subject when can be blank in a valid email. This will lump all of the bad emails together, which can be deleted enmasse. Now that you have everything in Mail.app, getting the emails to EagleFiler should be a simple process. Alternatively, you could do the clean-up just as easily in EagleFiler, but since EF has to index everything during import, it is probably a better idea to get the mail folder as clean as possible before importing.
This is all the fault of MailSteward. Who knows why a program dedicated to archiving email would generate bad email formats when exporting, but that is the way it is, and that is one reason why I’m consider EF for all of my long-term email archival in the future.