* Qmix fix.
-Current version of double Q learning is incorrect; it selects actions
at timestep t instead of t+1 when computing the t+1 Q value.
* Allow extra obs dict keys
* Move Q-value-computing replay code to own function
* Run the autoformatter
* use better terms in comments ("policy" network instead of "live" network)
* Validate that entropy coeff is not an integer
Passing an integer value for entropy coeff such as 0 raises an error somewhere inside the TF policy graph, so this checks to make sure the entropy coeff is a float.
* Cast to float instead
Also move this check after the negative value check
* custom action dist wip
* Test case for custom action dist
* ActionDistribution.get_parameter_shape_for_action_space pattern
* Edit exception message to also suggest using a custom action distribution
* Clean up ModelCatalog.get_action_dist
* Pass model config to ActionDistribution constructors
* Update custom action distribution test case
* Name fix
* Autoformatter
* parameter shape static methods for torch distributions
* Fix docstring
* Generalize fake array for graph initialization
* Fix action dist constructors
* Correct parameter shape static methods for multicategorical and gaussian
* Make suggested changes to custom action dist's
* Correct instances of not passing model config to action dist
* Autoformatter
* fix tuple distribution constructor
* bugfix
