We need an implementation of stochastic DCM

[hstrey]

The first step is implementing stochastic DCM with n linear mass models with thermal noise.
problems: We need to find a way to constrict our connection matrix to having negative eigenvalues.

Second step is to implement a version for arbitrary stochastic DEs.

For simpler systems, we could use MCMC; for more complicated ones, we need to use variational approaches.