Don't run prior and posterior predictive when calling fit

[michaelraczycki]

This PR contains the following changes:

sample_model() - removed from the fit() call, each sampling should be called directly and independently by it's corresponding function. Attributes handling moved to fit()
mypy fixes
docstrings updates and changes

---

📚 Documentation preview 📚: https://pymc-marketing--365.org.readthedocs.build/en/365/

[codecov]

Codecov Report

Merging #365 (fcd30c6) into main (4c9f4ca) will increase coverage by 0.88%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main     #365      +/-   ##
==========================================
+ Coverage   93.62%   94.51%   +0.88%     
==========================================
  Files          20       20              
  Lines        1664     1659       -5     
==========================================
+ Hits         1558     1568      +10     
+ Misses        106       91      -15     

Files Changed	Coverage Δ
pymc_marketing/clv/models/basic.py	97.87% <100.00%> (ø)
pymc_marketing/mmm/base.py	97.50% <100.00%> (ø)
pymc_marketing/mmm/delayed_saturated_mmm.py	100.00% <100.00%> (ø)
pymc_marketing/model_builder.py	83.06% <100.00%> (+7.52%)	⬆️

[michaelraczycki]

@ricardoV94 , @juanitorduz please let me know what you think about it, I think it might be first step on improving the time length our tests take to finish

[michaelraczycki]

merging this should close #364

[juanitorduz]

@ricardoV94 , @juanitorduz please let me know what you think about it, I think it might be first step on improving the time length our tests take to finish

Thanks! This was definitely needed! I still wanna avoid sampling for the plot tests :) I will do it on another PR