decide how to calculate reward_rate

[matchings]

mindscope_utilities.visual_behavior_ophys.data_formatting currently uses a method to calculate reward_rate from the SDK: https://github.com/AllenInstitute/AllenSDK/blob/master/allensdk/brain_observatory/behavior/trials_processing.py#L941

visual_behavior_analysis has a few different methods for calculating reward rate, and the one that is currently being used for most analyses is here: https://github.com/AllenInstitute/visual_behavior_analysis/blob/master/visual_behavior/data_access/loading.py#L483

We need to decide which of these we want to use for analysis that relies on mindscope_utilities

[matchings]

Note that the version of reward_rate calculated by the SDK takes the reward rate across change trials, then fills in each stimulus presentation in between changes with the value of reward_rate from the last change. This is probably not what we want to use.