Moderation mistakes are inevitable, and the consequences of a bad moderation decision can be as bad, or worse, than the effect of the message itself.
IMO, people should learn to defend themselves against trolls rather than helplessly relying on a nanny from above to do it. Unfortunately, using those platforms doesn't give us much of a choice in the matter.
That's a problem in some generic sense perhaps, but not with the application of this policy, which does not rely on a determination of “good faith criticism”, but instead of dehumanization.
Note that the examples of what is covered all are dehumanizing in a very narrow sense (stating directly that the target groups are vermin, viruses, etc.): https://blog.twitter.com/content/dam/blog-twitter/official/e...
Only if you "dehumanize" said religious people. All of the examples Twitter posted involved comparing the religious in question to animals (or viruses).
So, I doubt a tweet like "I don't think it's right to let the mother die because of fetus complications" would be removed, but something like "Those Christian pigs think a clump of cells are worthy of having human rights?! What kind of monkeys are leading them!" probably would.
I'm pretty sure you can criticize all of those positions without dehumanizing people on the basis of religion, so that conclusion seems unwarranted, and out of line with the examples of the kinds of posts that would be removed by the policy.