Document maximum likelihood estimation in gambit-logit

[tturocy]

The gambit-logit command line tool exposes undocumented features for computing MLE estimates of quantal response equilibria from observed date.

Add documentation on these features to the gambit-logit documentation, including, possibly, a tutorial for using the tool at the command line.

[tturocy]

This is dealt with in 7cacf5f. As our approach has evolved significantly in the years since this issue was created, the documentation is now enhanced in the user guide for pygambit. We will leave the ability to do the QRE fitting in the command-line tool as an undocumented feature (but listed in the help string), but doing this via pygambit is strongly preferred as it will provide much richer functionality.