Unify the initial data build and data updates so that everything is in a single dataframe?

[mikkokotila]

Right now there are two separate dataframes; pr_df and review_df due to them being of different resolution. This can be overcome by treating each row as an event, and duplicating the features from pr_df to the rows coming from review_df based on PR number (number). At least so far, I did not find any good reason for why to keep them separate.