Add a faster activation path

[avik-pal]

• ~sigmoid_fast fails on GPU as pointed out in the NNlib PR. Need to work around that~
• ~Specialize WrappedFunction on Broadcast.BroadcastFunction~ --> can't do this mutation will violate purity wrt input/output arrays

• #570 pulls out the easy part of this

Remaining Problems:

• [ ] Type Stability failure for DEQs
• [ ] NeuralPDE failure