Draft: B2OC: Improve selection of D*(2010)+ -> D0 pi+ (shifters please ignore)

The \delta m variable has much better resolution and is more efficient at rejecting background than m(D^0 \pi^+). Moving to \delta m should allow to reduce the rate, so that the requirements on the fdchi2 and direction angle can be removed. These have low efficiency due to the bad resolution of the D*+ decay vertex along the D*+ momentum direction (which is due to the low Q value of its decay).