Optimized reading large texts into RichEdit to be an O(n) order
algorythm instead of O(n^2) by removing extraneous conversions of
character offsets to run offsets.