Add test raw data token case in RLEDeompressor, fix off-by-one end check
data.mid(i, token)
reads to up to (i + token) (exclusive), but
maxIndex
is exclusive.
E.g. for the trivial {1, 0xff} maxIndex is 1, and reading token == 1
raw bytes at index i == 1
is valid (i has been incremented already).