Max's adaptive scheme

I don't know if he has a formal name for it yet and the name "Adaptive Transport Map" is a little all-encompassing so I'm not sure what to call it, but I'd love to get this up-and-running in C++. The high level steps according to Max are:

• Extract a bivariate marginal
• Use OT to sample each marginal to uniformly spaced samples between [0,1].
• Separate the samples in [0,1] x [0,1] into n x n bins. Count the samples in each bin. This gives you the observed histograms for the likelihood ratio test ($O_i$).
• If we use OT (rather than the Gaussian cdf I used in the slides), the expected histograms ($E_i$) are now uniform
• Calculate the G-test statistic
• Threshold it with the $\chi^2$ CDF

Obviously, this requires functionality for simple OT, histogramming, and $\chi^2$ CDF implementations.

Start by identifying where C++ transport-map functionality lives and whether the project already provides optimal transport, histogramming, or chi-square CDF utilities. Trace the proposed adaptive scheme from bivariate marginal extraction through bin counts, the G-test, and chi-square thresholding. Done means the scheme is implemented with a clear interface and validated for the stated workflow.