It seems to me that the complaint is "what used to fail with this error code now fails with a more annoying and scary error code." It's still a regression, but hardly one that's a huge cause for concern.

No. The complaint is that the error code is inaccurate and indistinguishable from correctly reported corruption. Moreover, it has an unnecessary disk check as a side-effect. This is more than a slight annoyance.

I would find it less suspicious if it didn't affect developers using tools that compete with The One Microsoft Way of Managing Source Code.

But my perception of Microsoft is admittedly biased.