initial_time_split & rolling_origin: Add `lag` parameter for predictions with step_lag

[mdancho84]

Problem:
Time series predictions with lagged variables aren't possible with intial_time_split() and rolling_origin() in certain situations.

Solution:
Add an overlap argument to offset the testing / assessment datasets.

A pull request is coming shortly.

[mdancho84]

Per discussion with @topepo, the appropriate parameter name is lag (instead of overlap).

[github-actions]

This issue has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex https://reprex.tidyverse.org) and link to this issue.