Nodes

[tomerarnon]

Noticed in #105 that Node really isn't required for maxsens.
It also should be quite possible to write it using our own approximate_affine_map or LazySet's overapproximate (probably). For example, we can write it as

function forward_layer(solver::MaxSens, L::Layer, input::Hyperrectangle)
    output = approximate_affine_map(L,  input)
    β = L.activation.(center(output))
    βmax = L.activation.(high(output)
    βmin = L.activation.(low(deviation))
    #etc.

[tomerarnon]

output = overapproximate(AffineMap(L.weights, input, L.bias)) is another option for this line. Any preference on which is clearer for the paper @changliuliu?

[changliuliu]

I like approximate_affine_map

[changliuliu]

This new implementation is very neat!

[changliuliu]

I like approximate_affine_map