checking singularity in first order approximation

Currently, we don't check for singularity in first order approximation when solving for static variables (is b10 singular?) or solving for shocks coefficient (is A_ singular?)

1. We should probably add a warning to stoch_simul (and ramsey_policy)
2. Should we care for estimation? Should expand the implicit prior to b10 and A_ non-singular?
3. If b10 is singular, the model has a problem: it is not possible to determine some static variable from the solution of the dynamic part of the model
4. The conditions under which A_ can be singular are mode difficult to determine.