lag_windows: Collects and embeds lagged windows of data as a vector.
In fnoorian/mltsp: Tools for machine-learning based time-series prediction

Order p determines size of the recent lags, and P and freq can be used to create seasonal lags.

1 2	lag_windows(x, p = 0, P = 0, freq = 1, shift = 0, no_peeking = TRUE, pstr = deparse(substitute(x)))

`x`	data input (can handle multi-column data, such as data.frames)
`p`	window length
`P`	seasonal window length
`freq`	seasonal period
`shift`	if the time-series is to be shifted
`no_peeking`	if TRUE, lags <= 0 (from now and future) are removed
`pstr`	name of datacolumn