Unintuitive handling of advancing simulations with internal times during playback

Most simulations have their own internal time which does not relate to real-world time elapsed.

The current implementation of the `OmniRunner` uses a `VariableIntervalGenerator` to call a `advance_by_seconds`function common to all Simulation types, passing the real world time elapsed from its last call (0.03 s by default).
As previous Simulation implementations have not properly implemented this function it has not been an issue. However, the recent implementation of the `UniverseSimulation` for interfacing with MDAnalysis presents a problem.
MDAnalysis simulations have an internal time step between simulation frames which has no direct relation to the real world time elapsed - i.e. there is no strong justification to tie the time elapsed between when `advance_by_seconds` was last called to how much the simulation should jump ahead. Rather it makes more sense to be able to set a given timestep which can be consistently called (and modified) irrespective of when the last call was made.

A possible solution to the issue is to subclass the `Simulation` protocol to a `RealTimeSimulation` and a `InternalSimulation` (names wip) which would expose either a `advance_by_seconds` or a `advance_by_nanoseconds` function.
The `OmniRunner` would then be able check the type of simulation and either advance the simulation by the real-world time elapsed if playing a `RealTimeSimulation` or a constant timestep if using `InternalSimulation`.

That said, the above suggestion may just be avoiding dealing with the root cause of the issue and a more substantive solution may be required.

Start with OmniRunner, the Simulation protocol, and UniverseSimulation, focusing on how advance_by_seconds currently receives elapsed time. Review the MDAnalysis integration and determine the agreed advancement model; done means playback can consistently advance simulations with internal time without relying on real-world elapsed time.

Written by the indexing model from the issue text.