Set reward-size on each trial

[micahwoodard]

Move reward size form task logic level to trial level. Allows to give fraction of reward for auto reward. Closes #85

At bonsai level, default reward is set to 2ul at initialization and during session, reward size is updated in calculate reward block

[micahwoodard]

@arjunsridhar12345 breaking change here. Reward size changes from task_logic.reward_size.right_valve_volume to trial_outcome.trial.reward_size.right

[micahwoodard]

@bruno-f-cruz I moved the calculating reward size to quiescent period. Moving to iti required setting it at the beginning of the session as well so made more sense to me to move into quiescent

[micahwoodard]

@bruno-f-cruz Task engine now includes pre trial select many which sets reward size for each trial

[bruno-f-cruz]

The distinctUntilChanged being inside the SelectMany doesnt really do anything since the sequence only fires once. The solution is to move it to the subject subscriber, ensuring you add it after filtering by side so you dont get false positives at the operator's input.

[bruno-f-cruz]

I pushed a small change to fix the DistinctUntilChanged logic. As it stood, it would almost always fire since it is added AFTER the merge (which is subject to values from BOTH branches). Also, HarpMessage (https://github.com/bonsai-rx/harp/blob/e7b5e0787c92e84a7eefdd98e797bc16036b5038/src/Bonsai.Harp/HarpMessage.cs#L8) are ref types, so equality defaults to ref equality not by value.

The solution for both is to add the DistinctUntilChanged operator after the value of EACH valve. This filters on the valve-specific stream with the added benefit that we are comparing value types (double).

[bruno-f-cruz]

Looks good to merge!