In python, custom factor without Jacobian

[goodgoodliving]

          > You create a custom function that computes `z-f(x)`. Then, in the `evaluateError` function of your customized factor, you can call the `numericalDerivative` function to compute the Jacobian numerically. Examples in `testNumericalDerivative.cpp` will help you get familiar with the function. 

@yetongumich
Hello, may I ask if it's possible to automatically solve for the Jacobian using GTSAM if I'm implementing a custom factor in Python, where I only know the error function and am not familiar with how to perform the Jacobian calculation? Are there relevant interfaces for this in GTSAM?

Originally posted by @goodgoodliving in #1547 (comment)