Resolve Issue In Linear Tests #2
Description
In test_cdnlgssm_filter_linear_TRegular.py, filtered means are compared between the analytic Kalman-derived filtering distribution and the (E(n))KF. The testing is a bit too strict and will always fail, but is never actually executed, as the error is caught for N_particles < 10^5, and tests are only run up to 10^4 particles. I'd propose using looser comparisons (say, allclose to 10^-3 relative error).