I’ve found that applying fades to 32-bit audio clips cuts everything within the selected range, but at any other it bit depth it works correctly – including 32-bit floating-point.
Thanks. Fixed in 3.1.1.