[chess] Nondeterministic qubit mapping failures

[losos0]

The code for mapping a circuit onto device qubits randomly alternates between success and failure for certain circuits. It seems like a better idea to return the same result each time.

This branch contains a test, test_entangled_qs_castle2 in quantum_chess/quantum_board_test.py, which fails to map qubits ~20% of the time.

I tagged it as a chess issue since AFAIK that's the only code that uses the circuit layout code, although the layout code is general-purpose.