Add alternative blocked on/off model.

[bcych]

Instead of Q matrices and update p_vec, write something where state is blocked or unblocked and do boltzman distribution. May have differences close to blocking temp.

[bcych]

States don't become blocked or unblocked - barriers do. This is a trickier problem than it initially seems.

[bcych]

Dropped to low priority