it would probably be a day or two of work for Pavel or me.
Actually, in rigid body refinement atoms do not move w.r.t. each other - the rigid body groups move as a whole, so I don't understand why one may need to use secondary structure restraints. The only restraints that make sense to use are those that would prevent rigid bodies from running one into another, but this is definitely more than a couple of days of work (for me, at least) -:) Simple geometry non-bonded restraint term is a possibility and easy-to-add-now option but not a good idea: one or two ideally coinciding atoms may throw the whole rigid body miles away from its original position (because non-bonded term grows exponentially as atoms get closer to each other). Possibly better is to use overlap functions (computed based on calculated density map, for example - so-called "soft non-bonded restraints") - but this is a little project for someone who's ready to jump into it -:) Pavel.