cv_assign_sets: Genrate cross validation sets
In rossklin/timetablr: Representation and basic operation for multiple time series.

Assign random indices to observations, useful for generating train/test/validation sets.

1 2	cv_assign_sets(tt, counts = NULL, props = NULL, sample.points = FALSE, cv.set.name = "cv.set", keep.all = FALSE)

`tt`	time.table to split
`counts`	vector giving number of observations in each set
`props`	vector giving proportion of observations in each set
`sample.points`	whether to sample individual observations rather than whole time series (defaults to sampling time series)
`cv.set.name`	name of column containing set labels
`keep.all`	whether to keep datapoints not assigned to a subset