4.1.16 [2018-12-19]
- Use forking instead of a socket cluster for parallel execution on UNIX-like systems for cvts(). This results in significantly faster execution and less memory usage, particularly when the FUN and FCFUN functions are very quick (e.g. snaive(), rwf(), stlm()), the time series is short, few cores are used, or few CV folds run.
- Faster cvts() examples.
- More tests for cvts().
- Fix spelling errors.