Long Data Sets not re-calculating correctly #3
Description
There is some issue with runs not recalculating on long data sets using xmR(). When the length of data is longer than about 20-30 data points and the data has high variance, it will sometimes miss the run or re-calculate at a different place than whats expected.
I think this is caused by the following scenario:
When bounds are initially calculated, there may be a high-variance run the the data which the bounds are recalculated to envelope. In this envelope, a lower short run is 'created' - a bit like an inverted U shape. The high-variance in the initial run means the function attempts to incorporate some of the high-variance points into the second re-calculation - which then 'creates' a new run and so on and so on recalculating bounds until it reaches a type of equilibrium.
I've tried using a conditional-while loop to test for runs, but couldn't get the test right to ensure it didn't go off into infinity. Instead, I've opted to just repeat some of the sub-functions inside xmR() to try to catch as many runs as is possible.
Another solution for this may be to add an argument for repetitions, but since MHC has a mostly stable use for xmR() I won't do this unless this bug actually crops up in use.