optimize and fix n-peg algorithm

[domino14]

n-peg for n > 1 is broken for non-bag-emptying plays. It often gives the right answer, but that's by accident.

We need to support at least one of the following modes:

When solving the peg recursively:

• when it is our turn again when the bag is not empty, solve the PEG recursively ("sure wins only")
• when it is our opponent's turn, solve the PEG on their side ("probable wins")

Also threads are not used optimally; when solving more complex plays and other threads are done they should help.