As it has been explained a couple of times, RSI, in its original form, uses what is sometimes called Wilders EMA for calculation (in values, it is exactly the same as SMMA (Smoothed Moving Average)). Replacing that average with some other type results in producing a completely different type of RSI than the original RSI is.
This version is using Alan Hull's moving average, and it uses it for two things:
It is a sort of "all Hull" RSI(oma) (it uses Hull average wherever average can be used or substituted. Also, this version is having the "divisor" parameter. That parameter, if set to 2.0 produces exactly the same Hull average as the original. Any value less than 2, "slows" down the Hull average, and any value greater than 2.0 "speeds" it up (but also adds overshooting - you shall notice those periods of overshooting when very sudden RSI changes of the "trend" happen - that is the consequence of being "too fast" and then some adjusting of the divisor parameter is a wise solution). So, one has to find some balance how to use that parameter and some experimenting is advised when it comes to its usage. All in all, Hull average produces sometimes "too fast" RSI values, but the results are interesting enough to play a bit with it since with right settings it seems to be a potential useful tool.