Expand interface, examples (both 1 new and improve existing)

A. Adds ability to do just the .hfo log to HFO.
B. Adds some .sh scripts for various example runs, including lengthy ones with logging other than .hfo disabled.
C. Adds getNumTeammates (and getNumOpponents) methods to avoid having to specify to the agent (either via arguments or via changing the source code) these numbers; makes doing checks of different combinations much easier. Updates hand_coded_defense_agent.cpp to use getNumTeammates instead of having a fixed value of 1 (cannot do this for getNumOpponents since code very much assumes 2 opponents).
D. Adds hand_coded_defense_agent.py - started with hand_coded_defense_agent.cpp, but more flexible regarding number of opponents and with more instrumentation.
E. Adds optional randomization of pass selection check order to high_level_custom_agent.py.
F. Adds ability to set (python) randomization seed to various python agents.
G. Adds ability to decrease difficulty of various python agents by setting --epsilon 0.#, causing random behavior (e.g., as per high_level_random_agent.py for high_level_custom_agent.py); also useful for comparison purposes, as per `random` in hand_coded_defense_agent.cpp.
H. Misc pylint and other cleanup of high_level_custom_agent.py, high_level_random_agent.py.

Update manual; avoid tabs in python files; get hand_coded_defense_agent working reliably

Added functions to get number of teammates and number of opponents - no longer have to specify for agents