Pybats-detection

- Fix for the `Monitoring` class when `y` has missing value
- Fix for the `Smooth` class with dynamic linear regression model, only the last values of the covariates were used to compute the smooth predictive distributions producing incorrect results

- Changed the parameter `change_var` to a dictionary `discount_factors` that received exceptional discount factor values according to the model block
- Improved class documentation
- Improved pybats_detection vignette

- Included the dynamic linear regression model into `Smooth`, `Intervention`, and `Monitoring` class
- Included `bilateral` and `prior_length` parameters for the `fit` method in `Monitoring` class
- Added new unit tests coveraging different scenarios
- Fixed the index of method `level_with_covariates` in `RandomDLM` class
- Renamed the arguments `type` and `distr` to `distr_type` and `distr_fam`, respectively in method `fit` from `Monitor` class

- Changed the names of model components. Now we are using the same name as `pybats`
- Included the sum of seasonality components in the smooth and filter posterior moments
- Fixed a index bug to compute the smooth distributions
- Included the `verbose` argument in `Monitor` class, allowing the user to control the monitor detection
- Allowed `None` values for the parameters of subbjective and noise type intervention, `h_shift`, `H_shift`, `a_star`, and `R_star` in `Intervention` class
- Added a new quick start guide vignette