Make CausalEstimand and Constraint Callables #86
Description
The CausalEstimand and Constraint classes currently have a public attribute do_with_samples, that is used to evaluate these components when passed in sample values.
To mask this somewhat ugly method of actually evaluating these components, we should make _do_with_samples private and implement a __call__ method for these two classes.
To be consistent with the Graph.model argument format, the argument to __call__ should be a dictionary mapping strings (node labels) to the sample values. This dictionary should then be expanded inside __call__ when referencing the (now hidden) _do_with_samples.
CausalProblem.lagrangian can then be updated to use the nicer syntax.