Okay, it seems I finally have the helium system under control - I've just tested a leak in the center engine reservoir with supplying the center engine from the pneumatic reservoir - works okay till MECO (and triggers the expected warning lights).
The system now has the correct pressure behavior, so only the highest pressure reservoir supplies the interconnect manifold and an engine only draws from the interconnect manifold if its own line pressure is lower than the interconnect manifold pressure (at least that's how I read the valve schematics...)
I suspect (since the helium flow rates are now constant during engine operations) that for long engine use (TAL and RTLS) it may now be required to let the failed engine supply the helium for the others.
I assume there would be cases when two systems fail that make it necessary t manually shutdown a good engine during the late ascent to make sure there's enough helium left - though I've never seen it documented, it would make sense to me..
Currently I assume that we start with 4700 psia and use 3300 psia during a nominal ascent for each engine (which gives some margin till the regulator pressure is underrun...). The pneumatic system is assumed to use 800 psia (so it has enough to compensate for a complete loss of one system).
If anyone has better data, let me know. Also, if anyone has data on how much additional helium the MPS propellant dump would consume - let me know (since the system is now fairly detailed, we can insert all sorts of behavior).
(Oh, and unless you use the property browser you can't try it yet because the 3d work for the switches isn't done yet...)