Why is this a particularly important ambiguity for people speaking to you to make explicit, compared to the thousands of other ambiguities inherent in the use of natural language?

There are thousands of ambiguities in natural language, but most of them don't have a connotation that I am a criminal in disguise. If it becomes accepted uncritically, some day it could have negative consequences for me.

How would e.g. a black person feel about a habit of inserting sentences like "this criminal was a normal black person" whenever a crime done by a black person is dicussed?

But also women have a selfish reason to care. Imagine that as a heterosexual woman you want to have a partner, and you want to minimize the risk of being abused. Changing the society and the legal system helps, but that is a very slow process. You also want to reduce the chance that you specifically will choose an abusive partner. So here is a specific man, and he looks attractive.. how can you estimate the probability of future abuse? Is there any evidence available?

Believing that "all men are abusers (when given a chance)" suggests that no evidence exists; there are no red flags you could detect to reduce your chance of future abuse.

I believe that this is false; a fallacy of grey. While there is no 100% algorithm to find a man who won't ever abuse you, there do exist some red flags, and by using them you can reduce the chance. An obvious example would be seeing the man behaving aggressively towards other people. (Some women ignore even this red flag.) I suspect that alcoholism may be another red flag; there are possibly more such red flags known. Discovering these red flags using statistics could be helpful.