Why are SWAP gates sometimes added during transpilation?

Why might a transpiler insert SWAP gates into a quantum circuit?

A

To reorder the classical measurement results so the output bitstring reads left to right

B

To reduce the circuit's depth by running two-qubit gates on more qubits in parallel

C

To copy a logical qubit's state onto a spare physical qubit before a noisy gate runs

D

To move logical qubit states so required two-qubit gates can occur on connected hardware qubits