Yeah, I haven't read the whole thread here, but I think I've got a solution for you (as mentioned by Teddy).
Here's what you do when you record:
SBD > Recorder 1
Mics > Recorder 2
I'm assuming that you are starting off with full batteries.
Later, you re-record one of the recordings onto the other device, taking care to start off again with full batteries and especially taking care to re-create the same thermal environment that you had when making the original recordings. (Crystal oscillators change their frequency of operation very slightly over temperature. It's not much, but the amount of temperature related drift over an hour's time can be significant.) So you set your recorders up exactly like they were, in the same bags, in the same orientation, with the same amount of charge on the batteries, etc... so that you get as close a possible to the original recording conditions. However, instead of recording a live source, you re-record one of the recordings.
Recorder 1 (playing the original SBD recording) > Recorder 2 (recording)
Recorder 1 will play back the SBD recording at exactly the rate that it was originally recorded. Recorder 2 records that signal as if it were live. The net effect is that you have recorded both the mic source AND the SBD source on the same recorder. My experience has been that you can hold a 2 ms/hour skew between the two sources when you process it this way. Yes, it's an additional a/d to d/a process on one of the recordings, but it's still better than trying to stretch a whole bunch of small sections of one of the recordings to keep it in synch with the other recording. If you're careful, you shouldn't hurt the S/N by more than about 1 dB or so. (I'm assuming that you paid attention to your levels and were careful to use the whole input dynamic range. If you do this with a quiet recording, you'll lose more S/N, so be sure to keep your levels up when doing your original recording.)
One time (before I started re-recording one source like I've described above), I synchronized two sources by stretching 2 second chunks of one source every 6 seconds of the recording. That's 600 edits per hour. The show was over 2 hours long. It took me hours and hours of editing to accomplish the alignment and I had to design a spreadsheet to calculate all of the edit points, working from the end of the show towards the beginning. I thought I was never going to get that thing done. Actually that show will be posted on Archive.org sometime in the next month or so. It was the Hank III/Assjack show from Carbondale, Illinois a few years ago. I've sent the flacs to a buddy of mine who has promised to upload it to the archive. Anyway, that method worked but it was waaaayyyy too time consuming. Ever since then, I've just re-recorded one source onto the other device, mixed and tracked. Easy. Quick. Done! And it doesn't produce any noticeable degradation in the sound quality if you're careful with your levels.
If you don't like the idea of re-recording, you can always get a multitrack recorder. My next project is to share the master clock between two UA-5's, essentially giving me a poor man's R4 or 744.