A few other notes on the output. There will always be two lines at each intersection, with the Road Names in sorted and reverse sorted orders. If only two road names meet at an intersection and you only want one line, you could select based on the sort order and either delete what you don't want or export what you do want. (I would export and leave the result unedited to be able to go back to it as a reference during clean-up).
At intersections with 3 or more names there will be more lines. Also the names that are opposite each other across an intersection with more than 2 names will place the crosswalk line in the opposing named road's location. So some process to identify these intersections and switch the attributes would be needed. Also, at 4 way named intersections you would need to deal with revising the field structure to handle more than 2 names if you want a single line at each of its Crosswalks, since at least 3 names would associate with each crosswalk.
So there will probably still be about 10 to 20 percent of the segments that need manual clean-up. I find that is typical for virtually every first-pass, broad scoped geoprocessing operation.