I think your losses are pessimistic. Even using about double the flow rate we were talking about to keep inlet temperature down (1GPM) you shouldn't need 50mH2O of head. A common 360mm radiator like the Swiftech MCR360-QP has a pressure drop of ~0.2mH2O at 1GPM. Even three of those in series would only be 0.6m of pressure drop.
I don't know how those blocks are internally constructed, but most common CPU waterblocks are 0.5-2mH2O @ 1GPM. If those two blocks in the C1 are close to the 10mH2O at that speed you would need in order to get a 50m total drop, I would be extremely surprised and they would be a terrible design.
Do remember that the radiators only have essentially 3 180 degree turns and large flow channels in order to slow the fluid down. This works because of the extreme surface area. The waterblocks on the other hand (of which there are 10), has a small diameter channel and the flow snakes forwards and back 4x [4x 180 degree bends]. Each barb, fitting and slight bend then adds additional minor losses. The figures you're quoting are also for 9.5mm ID tubing, where as we're using 6mm.
Even if you pretend for a second there are zero minor losses through the entire system, just look at the major losses. There is about 1.2 metres of path in each block, so 12m of path just from them. The radiator has another metre internally, and you've got all the small bits of tubing and then the larger lengths of tubing. Even taking it down to 15m, that's a major head loss of 22.5m.
And yes, this style of waterblock is MUCH more restrictive than a radiator as you have to snake the channel because you have no internal fins for surface area.
I never said it wasn't more restrictive than a radiator, just that it would be pretty terrible if ten of them in series had 50mH2O (71PSI) pressure drop at 1GPM.
Since Bitmain is selling these without matching hardware, it would actually be really handy if they could make flowrate vs pressure drop chart for them, similar to the one in this datasheet.