In the splitters:

The BNL_magnet directory shows some Halbach quads that may be of use for the splitter.  Dipole versions of these can also easily be made.  With these, a ~10% electromagnet would be placed after the Halbach magnet.  This is just to avoid having overly-large electromagnets and their iron yokes in the splitter.

In the main FFAG:

I'm finding it quite hard to fit corrector coils inside the main magnets.  Consider stepper motors geared down to low displacements less than 1mm range (or even some piezo mover).  These can provide both dipole directions via displacement and the skew quad via rotation.  The only thing left is the normal quad adjustment.  Holger has some ideas about moving iron blocks near the magnet to control this.

So it's possible to get decent correction using about 4 stepper motors (~$80) on each magnet.  With appropriate diagnostics, these steppers could compensate for initial misalignments as well as temperature variations and other slow changes like radiation damage or slow mechanical shifts.

The only thing these can't do are fast vibrations of a few Hz and upwards.  These are typically 1um level or better, but it would be nice to know Cornell's vibration environment there.  The 1um level is 0.01% level so a hundred times less than we were thinking before.  Electromagnetic correctors at this level should be easy and fit well within the magnet bore (and work with our 10A/20V power supplies).