test_AD "normal" kernels vs MOKernels

https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/pull/263 by @david-vicente introduced separate methods for test_AD for MOKernel - I'm concerned this means that we won't be running the same set of tests for MOKernels.

Maybe that's how it has to be, but then it seems like the type hierarchy isn't quite right - we should obey Liskov substitution principle, right?

This to me seems like a case where overloading the same method isn't really the right thing to do (because it suggests they do the same thing - it's a kernel AD check in either case - but they don't actually seem to check the same things)... now that i've got y'all in a thread, what is your opinion on these ?

Originally posted by @st-- in https://github.com/JuliaGaussianProcesses/KernelFunctions.jl/pull/414#r771854325

Start by reading PR 263 and the discussion quoted from PR 414, then inspect the separate test_AD methods for normal kernels and MOKernel. Determine whether both kernel types should receive equivalent checks and whether the current type hierarchy supports that substitution. Done means the testing and hierarchy approach is agreed and the resulting behavior is covered consistently.