MATSim Code

a couple of changes to msa score averaging (making code consistent with what we want to write in the book)

add capability to identify toll avoiders

changes in kai part of SA playground

fixing a problem caused by using link.getCapacity() rather than link.getFlowCapacity() [arrghhh]

unfortunately, lots of collisions in QLink/QLane area that needed to be resolved manually

legModeRouteDistribution data based on events file.

legModeRouteDistribution data is now written from events not from outputPlans (executed plans are not same as experienced plans) and some more editing

marginalCongestionEvent time corrections, as per rev_30447

trying to fix random errors with PT simulation on machines with more than 1 CPU. Although the data structures are accessed in a thread safe manner (all add happen in one thread at one time, and afterwards all read and removes happen in other threads at another time), I suspect that due to local caching allowing by the Java memory model, some inconsistencies can come up. Those trying to solve this issue by using data structures that should guarantee that modifications are always visible over multiple threads.