Incorrect use Lux.ActivationFunction

[avik-pal]

The updated docs have used ActivationFunction in https://diffeqflux.sciml.ai/dev/examples/neural_ode/ (and possibly other places). Note that the semantics of ActivationFunction is different from the intended use case -- See http://lux.csail.mit.edu/stable/api/layers/#Lux.WrappedFunction vs http://lux.csail.mit.edu/stable/api/layers/#Lux.ActivationFunction (the intention of having this layer was to allow a very rudimentary implementation of layer fusion, which I might still implement later on)

Additionally, this should not be done, since http://lux.csail.mit.edu/stable/api/layers/#Lux.Chain will automatically do it for the user (and IMO makes code less clunky)

[Abhishek-1Bhatt]

Should we switch to WrappedFunction in the docs then?

[avik-pal]

No no. Just remove any wrapper, Chain will do it on its own.

[Abhishek-1Bhatt]

Oh, yeah it returns it as a WrappedFunction by default