Quantum Chess: use calibration metrics for better qubit picking

We should explore using calibration metrics (https://quantumai.google/cirq/google/calibration) as a way of picking better qubits for quantum chess (i.e. picking qubits with lower sq/sqrt-iswap/readout errors).

See also quantumlib/unitary#51 and quantumlib/ReCirq#127 for similar issues.