pair_style Brownian with log terms not suitable for separations greater than the radius

Steve and others,

This is just a note that I would suggest putting into documentation. If pair_brownian is given a cutoff greater than the radius of the particle the log terms in the lubrication forces (a_sq, sh, and pu) become negative. This creates problems when trying to find the structure of the random forcing (which necessitates taking a sqrt), namely nan forces.

In general I would caution against using such large cutoffs for the force even in the pure lubrication case, e.g. in the high Pe number limit, as it switches the direction of the forcing.

Another option would be to only retain the 1/x a_sq term beyond that limit.

I see that it has already been included in pair_style lubricate documentation (dumb me). Repetition in Brownian, I think, would still be useful - since it won’t work at all there.