added check_error to atsit5

[rusandris]

The check_error for DEIntegrator doesn't work for the simple integrator types because they have different fields. I've added a simple function to handle atsit5 which is the default solver in DynamicalSystems.
See: Issue 184 at JuliaDynamics/DynamicalSystems.jl

[rusandris]

I'm not sure if this is the right place to add this function. Would this check_error defined for simple integrator types and the one from SciMLBase collide?
Also, should I open an issue for this instead?

[ChrisRackauckas]

Why not just use Tsit5? The purpose of the simplified version is that it cuts out pieces like this, and the error check is the vast majority of the overhead cut out. If this is added... we're basically back at Tsit5?

[rusandris]

You mean changing the default to Tsit5? I don't think it would be a good idea to change the default just because of this. But yes, you are right, we thought that the error checking should work the same way for all types of integrators, but the best option indeed would be if we leave the simple integrators as they are, and use the error checking only in the other cases in our functions that use integrators.