modify LMOptimizer's parameters

[tzamos]

Hi,
I wanted to modify the parameters of LMOptimizer, in particular the "num_steps" of LM optimization. I noticed that when calling the constructor of GeoCalib class with the weights as parameter, the constructor from extract.py is called. there in the LMOptimizer is initialized with no custom parameters. Thus is it not possible to modify the parameters of LM through this constructor. It was easy to hack it, by adding a parameter num_steps on either the constructor or the calibrate function, and modify it as: self.model.optimizer.num_steps = num_steps.

Thanks!