Media transforms prior sampling & curve plotting

[wd60622]

Following #632, it's easy to build in support for adstock or saturation level sampling and plotting to assess the priors and posteriors independent of a larger models.

For instance, the workflow might be:

from pymc_marketing.mmm import MichaelisMentenSaturation

# Analysis outside of the model
saturation = MichaelisMentenSaturation()
prior: az.Dataset = saturation.sample_prior()
saturation.plot_curve(parameters=prior) 
# Produces saturation curve along some default access

Which could be helpful for analysis of the posterior as well.
For instance,

mmm = MMM(..., saturation=saturation, ...)
mmm.fit(X, y)

# Could be external or used internally as well.
mmm.saturation.plot_curve(parameters=mmm.fit_result)

[juanitorduz]

Nice 🚀 !