Error in timesplitting for tau = 0 cases

[shyams2]

In: f90bc8b

For tau = 0 cases, the splitting has not been carried out correctly, and is just applying f = f0 at the end of the FVM step in a single timestep.

Proposed Fix:
Considering strang splitting:

1. Solve df/dt + A_q df/dq + A_p df/dp = 0 for dt/2
2. Apply f = f0
3. Solve df/dt + A_q df/dq + A_p df/dp = 0 for another dt/2