BuildModelFromDAG class for causal components (#1950)

* BuildModelFromDAG class for causal components

Introduces the BuildModelFromDAG class to build PyMC probabilistic models directly from a causal DAG and tabular data, with validation for dims, coords, and priors. Adds comprehensive tests for parsing, validation, error handling, and model construction, ensuring robust integration and usage.

* Lazy imports because this works only with causal extras in pymc marketing

* Apply William feedback

* Adding new tests

* Solving test issue

* Adding intercept

* Apply changes by Juan!

Author: cetagostini