cusum_control_df is required so time based small-multiples are easier to produce

[johnmackintosh]

• There's a need for a cusum_control function that can handle a dataframe / tibble / data.table ( my original expectation was I would use this infrequently for standalone plots, so I only ever expected to pass in a single vector, rather than need a grouped df)

Arguments:

• df name
• x date or observation / identifier vector
• y (vector to calculate limits against)
• grouping_var

I need to think about whether I go all in on dplyr for this package, rather than data.table. I have a grouped_modify_join function but group_modify is still experimental? So don't want to bet the farm on it in case it gets pulled.

Using dplyr would make non standard evaluation much easier and rlang / ggplot2 are already imports so probably makes most sense. data.table's NSE is changing and I may not have time to implement it.