Improve efficiency of _scan method in TokenizedLine class

[mbuhidar]

Currently the method cycles through character maps for each token in each line. Not necessary based on max size of listings vs modern compute, but would like to improve the approach to this.