Implement logcdf method for discrete distributions (#4387)

* Implement logcdf method for discrete distributions

* Add release note

* Small reformatting

* Add more comprehensive bound check for impossible parameters

* Fix bounds for logcdf of DiscreteUniform

* Add safe values/ parameters for `BetaBinomial`, `Poisson`, and `HyperGeometric` to avoid errors when evaluating negative logcdfs.

* Change `check_selfconsistency_discrete_logcdf` to work with negative domains, such as the DiscreteUniform.

* Raise TypeError in logcdf methods that only accept scalar values.
Fix docstring formatting.

* Small change to logcdf methods of `DiscreteWeibull` and `ZeroInflated` making use of `tt.log1p` and `logaddexp`.
More informative comment on workaround for `Poisson.logcdf`.

Author: ricardoV94

RELEASE-NOTES.md

@@ -19,6 +19,7 @@ It also brings some dreadfully awaited fixes, so be sure to go through the chang
 - Option to set `check_bounds=False` when instantiating `pymc3.Model()`. This turns off bounds checks that ensure that input parameters of distributions are valid. For correctly specified models, this is unneccessary as all parameters get automatically transformed so that all values are valid. Turning this off should lead to faster sampling (see [#4377](https://github.com/pymc-devs/pymc3/pull/4377)).
 - `OrderedProbit` distribution added (see [#4232](https://github.com/pymc-devs/pymc3/pull/4232)).
 - `plot_posterior_predictive_glm` now works with `arviz.InferenceData` as well (see [#4234](https://github.com/pymc-devs/pymc3/pull/4234))
+- Add `logcdf` method to all univariate discrete distributions (see [#4387](https://github.com/pymc-devs/pymc3/pull/4387)).
 
 ### Maintenance
 - Fixed bug whereby partial traces returns after keyboard interrupt during parallel sampling had fewer draws than would've been available [#4318](https://github.com/pymc-devs/pymc3/pull/4318)