subset.time.table: Subset a time table
In rossklin/timetablr: Representation and basic operation for multiple time series.

Subset a time table rows and columns by expression or indices, preserving time.table structure.

1
2
3

## S3 method for class 'time.table'
subset(tt, expr = NULL, vars = NULL, index = NULL,
  times = NULL, preserve.frequency = NULL)

`tt`	time.table to subset
`expr`	Expression evaluated within tt to subset the rows of tt
`vars`	Non-index/time columns to subset
`index`	data.table/data.frame containing indices or indices and times to subset
`times`	times within each index to subset (to not specify if index already contains times to extract!)
`preserve.frequency`	whether to preserve (or reguess) frequency information after subsetting

Note that for now vars *must* be a collection of strings.

One can specify more than one of expr, vars, and index/time, in which case the time.table is first subset by index/time, then by expr and finally by vars, meaning index/expr can rely on variables removed after the subset. Note that this can result in unexpected behaviour if expr evaluates to a vector of row numbers.

Furthermore both vars and expr can mention extra columns in index/time (merged in the intermediate data.table in the index step).

If a column name exists in both tt and index/time then the name will referr to the corresponding column of tt *if* the column is to be kept (i.e. it is in vars), else it will refer to the merged value from index/time.