Likely bug in xLASWP for less common increments

[markusm73]

I was looking into how to correctly do parameter checks for bindings to the xLASWP functions. After carefully reading both the man page and the code, it seems obvious that not only is the manual inconsistent with what the function does, the behavior of the function also seems completely unintuitive for increments other than 1 or -1.

If, for example, the increment is larger than 1, the function will access "K2 - K1 + 1" IPIV elements starting at K1 and separated by the increment. This means that even if K2 is well inside the bounds of IPIV, the function may go past the last element. This strange behavior seems in conflict with the manual and is probably not intended.