Try different preconditioners for the sparse LLS

[jlperla]

For the solution to the stationary_distribution_discretized_univariate.m and others, try to see if other preconditioning strategies work well for solving the LLS problem. For example, look at http://epubs.siam.org/doi/pdf/10.1137/140975358 and http://www2.cs.cas.cz/~tuma/ps/bru_mas_marin_tuma_2014.pdf for some references.