User weighting

[michaelzevin]

We need to incorporate a more reasonable user weighting for each user's contribution to the posterior probability on a given image. This was implemented on the "retirement" branch by using the "weight_users.py" function, which gives classifications made by users that are in levels 1-3 0.5x the weight of the machine, users on level 4 1.0x the weight of the machine, and users on level 5 2.0x the weight of the machine. This will work for the time being, though I am working on a more flexible way of incorporating weights.