That is the same conclusion I came to with the LN2 cooled Bitter magnets. Add one more. Standard Cu tube is limited to about 500 psi. If the magnets are in series you will go well above that. You have to balance electrical resistance against pressure drop. With any cooling method your outlet pressure must be well above the boiling point at the outlet temp to allow for local hot spots. That means a flow restrictor on the outlet to keep the pressure up.There are further tradeoffs regarding magnet electrical/cooling arrangements. I favor making each magnet independent rather than connecting all in series. I have three reasons for this. The first is, you want to make the cooling path as short as possible. The second is, it would be nice to eliminate those short interconnecting stubs on WB6, right in the spot where they wanted to eliminate the "funny cusp". The third reason is it would be convenient to make all the magnets identical to avoid the hassle of the interconnection bits, and to allow very easy replacement of individual magnets.
The downside is, each magnet must be on a big feedthru plate, on, I think, four insulated standoffs of substantial size. Two carry magnet current and water, two are just support. These will interfere with outside electron circulation, but I'm betting that is less trouble than the interconnects(relatively low electron density outside the coils), an using four insulators allows them to be place wherever the models suggest the density is lowest.
I like LN2 because the Cu resistance goes down allowing much smaller magnet supplies. I like the Bitter design (even though it is not volumetrically efficient) because of its mechanical strength and because it is just a series of punchings (perhaps gold or silver plated at the overlaps) and mylar insulators .0005" thick maybe less. I figured a volume fill (square form factor) of 90% with .001" mylar. I could probably go to 99% with careful handling of the mylar (.0001 thick).
Cu has a pretty high temp. coeff. of resistance making water cooling problematic. Of course it becomes self limiting in terms of current so that helps. Unless you have constant current supplies. Which is what you want for experimental purposes since every thing must balance - well depth, drive voltage, B field.
My plan was to use the "dead space" in the coil forms as flow headers and for running D-D gas injection tubes.
Some design studies (better than my +/- 20% BOE calculations) would need to be done to choose between the alternatives.
Of course LN2 limits you to 30 minutes to an hour per day of run time. Good enough in the beginning. In fact at the beginning a few seconds of run time would be more than enough. Given the usual problems.
Another thing LN2 gets you is high insulation resistance without having to worry about contaminates.