In representative metaphor, the words for “foot,” “advance,” or “withdraw” need not each acquire separate figurative dictionary senses. The effect belongs to their coordinated relationship. The words build a literal picture, while that completed picture is borrowed for another context.
The arrangement matters as much as the components. Counting words is a different task from identifying the unit whose meaning has been transferred.
Accordingly, to say that the whole utterance is borrowed is not to claim that every constituent has become an independent single-word metaphor. The classification concerns the composite scene as a unit.