"When doing digital transfers between a big digital multitrack setup and something like a Digi001 can one get by without a synchronizer? If not, what's the minimum synchronizer I can get away with?"
The answer, as is often the case, depends. First, if the number of tracks you need to transfer can be recorded simultaneously into the Digi001 then a synchronizer may not be needed. Who cares what the time code values are as long as you get the audio in there, right? Not necessarily. If you need to later transfer them back to the multitrack and keep them in sync with material already there then you'll need a synchronizer.
If the number of tracks to transfer is beyond what can be brought into the Digi001 at one time (one tape pass) then you need a synchronizer anyway. So odds are you probably need some kind of synchronization. The minimum you can get away with depends on what sort of timing information is available from the multitrack and how picky you are about timing between your tracks. On each transfer all of the tracks you move will stay in time with one another, but when they are combined with other tracks that were either moved on another pass or not moved at all there is potential for them to be off a little bit (or a lot, depending upon how you do it). The first thing to understand is that while doing a digital transfer the clocks of the two systems will have to be tied together to avoid clicks and pops. This also accomplishes the important job of keeping the two machines 'resolved' to each other from a synchronization point of view. Simply, it keeps their timing the same so they don't drift relative to one another. Now all you really need to do is make them start at the same time.
If you can pull SMPTE time code off of the multitrack system you can use a SMPTE capable MIDI interface (which isn't very expensive) to achieve a start time for the Digi system. The SMPTE interface converts the LTC to MTC, which is what Pro Tools (or any MIDI sequencer) needs in order to start playing at the right time. This solution is adequate for many users, however, the start accuracy is pretty much limited to a quarter of a frame. That means it may start as much as a quarter of a frame off (sometimes more). Depending upon the frame rate of your time code one frame is somewhere around 1/30 of a second, so a quarter of a frame is not much of an error, but in the world of digital it's around 350 samples - not wholly insignificant, it just depends on what you are doing and how picky you are. It's certainly enough to shift the stereo image of a pair of stereo tracks so you should be careful when you decide which tracks to transfer on each pass. Try to keep all the drum tracks together, for example. But don't forget you can easily slide tracks back and forth in Pro Tools if you want to change the timing.
If you require tighter timing than this you should get an interface the resolves the SMPTE time code with the clock rate of the master machine and gives the slave machine not only a resolved clock signal, but also MTC that is resolved with that clock. Your start time would then be nearly (but usually not quite) sample accurate. Digidesign's USD and MotU's DTP are both great for this. There is also a lot to be said for the quality of the master clock that drives the system in the first place. Many people believe the clock quality (stability) has a profound impact on the overall recording quality, but that's another article.
