stop_after = 0 in pygambit

[rahulsavani]

For methods like enumpoly_solve, stop_after = 0 is used at the C++ level but also works in pygambit, being equivalent to stop_after = None. This will be confusing for a user and should probably be removed from pygambit.

[tturocy]

Here's a suggested line to a solution.

The underlying problem is in C++ we conventionally use int stopAfter = 0 to mean "no limit" - because back in the day there was no way to have a "null" argument.

Instead, what we should do is to have C++ methods take a std::optional<size_t>. That automatically guards against negative numbers, and 0 should be checked and rejected as a bad argument. If the argument is equal to nullopt_t then that means no limit.

Then in Python, we should do similarly; stop_after should always default to None if it does not. We'll always check for stop_after <= 0 as an invalid argument, and then construct the appropriate value to pass to the C++.