TravelerList.clinched_by: list -> set

[yakra]

Converting TravelerList.clinched_by from a list to a set yielded a slight speed increase in C++.
This is worth trying out in Python.

[yakra]

C++: Just rewrote Route::clinched_by_traveler to take advantage of the fact that TravelerList::clinched_by is a set rather than a list now.
(That particular bit won't be required in Python.).
Creating per-traveler stats logs and augmenting data went from 28.3 s to 9.7, speeding up by a factor of 2.9. Not bad. I've considered this the most interminably long part of the program for a while now; this makes the wait a little more tolerable.

Where else did I see speed increases in C++? Maybe it was concurrency augmentation, but I'm not 100% sure.
ToDo: make a list of the tasks that are likely to be affected, and check times.

[yakra]

0ede2eb

task	before	after
llst processing	35.9	34.5
concurrency augments	13.0	11.6
Creating per-traveler stats log entries	77.7	57.9

List processing should in theory be a wee bit slower. The increased speed is probably due to disk caching.