Limit time step based on control system timescales in LTS

[wthrowe]

Quoting @markscheel from elsewhere:

The thing I am specifically worried about is that the control system has some damping time T, but updates fail to be done on a timescale of T/3 , which will happen if the timestep is larger than T/3. In that case I think the control system will go unstable. Presumably with LTS the timestep that matters is the timestep in the elements that cover AhA (the BH that goes bad)

I don't think we currently have any way to determine what elements are involved in a control system measurement, so we don't know where in the domain to impose a limit on the time step. This might tie into #4765.