generate.stat.dt: Construct a statistic distribution generated by permuting...
In mattmoo/SteppedWedgeAnalysis: Simulation and Analysis of Stepped Wedge Data

Statistic is Wilcoxon-Mann-Whitney at the moment.

generate.stat.dt(
  data.dt,
  cluster.dt,
  sequence.dt,
  comparison.within = c("cluster", "period")[1],
  hypothetical.data.dt = NULL,
  perm.dt = NULL,
  max.r = 1000,
  sort.input = T,
  exclude.transition = T,
  stat.func = c(test.wilcox.dt, test.f.effect.dt)[[1]],
  progress.bar = T
)

`data.dt`	data.table with columns participant, cluster, time, and outcome. Outcome should be continuous.
`cluster.dt`	data.table with the correspondence between cluster and sequence, with columns cluster and sequence.
`sequence.dt`	data.table with information about the sequences, with columns sequence, transition.time, and intervention.time.
`comparison.within`	Will comparisons be within cluster (i.e. between period), or within period (i.e. within cluster).
`hypothetical.data.dt`	Precalculated table of how outcomes would be assigned to groups if clusters were in each different sequence, will be generated if NULL (Default: NULL)
`perm.dt`	Precalculated table of how to permute clusters to sequences, with first column being clusters, and then one column for each permutation after that.
`max.r`	How many permutations?
`sort.input`	Will sort the input by outcome (by reference), which slightly speeds ranking
`exclude.transition`	boolean, should the result exclude data points from the transition period? (Default = T)
`stat.func`	What statistic will be used, there is a wilcox and f.effect as I'm writing this. (Default: Wilcox)
`progress.bar`	Display a progress bar. A little bit of overhead. Completion times will be echoed regardless.
`...`	Passed on to the test