Use Tuple for parameters instead of Array

[sathvikbhagavan]

This is to avoid a performance warning:

https://github.com/SciML/SciMLBase.jl/blob/master/src/performance_warnings.jl#L12

As the parameters of the FMU can be of different types

[ThummeTo]

Dear @sathvikbhagavan,

I need to think about this... because parameters may change between simulation runs... or in theory also during simulation.
on the other hand, one could build a new ODEProblem for every run (as it its currently implemented)... I will check if this would work.

[CasBex]

I know I'm late to the party, but if I look at the SciMLBase code then the problems is not with having arrays SomeArray but with having arrays SomeArray{SomeAbstractType}. While having an empty tuple as default argument is certainly ok, I would still dispatch on p::Any for better flexibility (this is what DifferentialEquations.jl does I think). If users decide to use setupODEProblem manually for whatever reason, they will get the performance warning anyway and can decide for themselves to improve their code or not. If they don't everything is fine as well.

[ThummeTo]

Will try and apply ;)