causalglm supports categorical and continuous treatments

The functions npglm and msmglm now support categorical treatments through the treatment_level and control_level arguments.
Also, causalglm now allows reusing previous machine-learning fits by passing in a previously obtained causalglm fit object into the data argument. This allows you to compute estimates based on different estimands and different formulas using the same machine-learning fit, which saves a lot of computation time. Check the vignette or readme to see how this works.

A new function contglm has been added with a focus on continuous treatments. Currently, only the CATE estimand is supported.