In different places, not an issue. It's only when penicillin is directly applied to a bacterial colony is there any chance for bacterial evolution to take place.

> Then practically overnight the bacteria decided they needed to evolve

You mean, when humans applied penicillin to bacteria on a large scale for the first time? That changed the bacteria's environment, most died, except those naturally resistant. It's classic natural selection.

> I don't think evolution always works according to the popular "and then the next giraffe's neck was 1mm longer than its parents'" gradual change model.

But one can argue that evolution almost never works, on the ground that the vast majority of mutations aren't adaptive. But the argument misses the point that some tiny fraction of the mutations become the entire future species because of increased reproductive fitness.

Also, the "gradual change" you describe normally arises because of the odd beneficial mutation, which, apart from being very improbable, might require many thousands of years to manifest itself.