Update prompt_response_tests.md to use safety example instead of gender bias

[brianwgoldman]

ModelGauge can definitely be used in this way, but it isn't our flagship usage. The example also relies on completion instead of chat, which isn't as core to what we do.

I think this is a low priority change, but wanted to document it as something we should do eventually.