I agree, and Chesterton agrees. He's very explicit in saying that if you can accurately identify the reason a fence is there and that reason is no longer applicable or was never a very good one, you are justified in removing it. The point isn't to never tear down a fence, it's to question your impulse to tear down every fence the instant you see one.

It's very much the same idea as "don't start ripping out and refactoring 'legacy' code on your first day on the job." You may actually have inherited a completely garbage code base, but you may also simply not understand the constraints that dictated the architecture you've been given. Once you understand those constraints, refactor away!

With regards to your example: Dietary laws may well be outdated, but modern Jews may also argue that setting themselves apart is still a very worthwhile reason to avoid pork—it's probably part of the reason why they managed to remain a separate culture even without a proper homeland, and remaining a distinct culture is valuable to them.