ENH: OrderedModel - make params_table for linear thresholds available

[josef-pkt]

two statsexchange question why threshold parameters in OrderedModel differ from SPSS and R.

https://stats.stackexchange.com/questions/592680/why-do-statsmodels-python-and-mass-r-output-different-intercept-values-for-m

we already have helper methods for converting threshold params

We could also make standard errors available so that we can create a params_table with linear coefficients instead of the current parameterization.
This will require NonlinearDeltaCov because of multiple correlated parameters and nonlinear functions.

[brienna]

+1