Improve Default Interval when using simulateCS

[halentin]

Currently, if the user doesnt pass saveat or dt to simulateCS, we fall back to dt=1e-3, which is a decent default value for short Simulations, but leads to really long simulation times for longer timespans (t_stop-t_start>10). When not provided with any arguments, i would expect the simulation environment to pick a more sensible default value. Both Dymola and FMPy do this by picking a dt so that the number of saved values is 500 (Dymola) or 500-1000 (FMPy). This PR copies the behaviour from FMPy.

As we dont have any other Files apart from sim.jl, i added auto_interval in there. Maybe it would make sense to put it in a new file like utils.jl or helpers.jl?

This could possibly be considered breaking because it somewhat changes the default behaviour?

[halentin]

Note, that currently tests dont pass, because of an older version of BoundaryValueDiffEq failing precompilation, and the FMISensitivity-Test-Dependency blocking from updating to a newer version (that actually precompiles). The CS-Tests (they dont require FMISensitivity) are passing.

[ThummeTo]

Can you please add a line comment that links to the file within FMPy if code is reused?