The same reason they are not thrilled about support in general: Manual labor is expensive, and at the scale that Apple or Google operate, the amount of abuse and unwarranted complaints you receive at high traffic, low entry barrier, outward facing services is absolutely insane and can not realistically be shouldered without heavy reliance on tools and templates.
But how can it be hard to make a bot produce the input it flagged?
I can see that if you are talking about humans asking them to compose detailed and specific reports has costs. But we're talking about machines, not humans.
Basically all anyone is asking for is a stack trace so they can identify and debug the issue. I don't understand why that's unreasonable for a bot.
It's like a complier that aborts with nothing more than "syntax error". Most people know compilers can be much more helpful than that.
