Deep learning model estimation

[sbenthall]

Once we have a general model language/object in HARK, and a robust monte carlo simulator (see #1296) it would be good to have a suite of estimation methods.

@JLDC has developed a "Deep Simulated Moments" technique (presented at DSE 2023) and is working on a Julia implementation.

It would be nice if HARK had a Python port of this method.

https://www.jldc.ch/publication/2023_deep-simulated-moments/ "Constructing Efficient Simulated Moments Using Temporal Convolutional Networks" https://github.com/JLDC/DeepSimulatedMoments.jl