That's a peeve of mine. I agree with the OP and the comments, as they tend to focus on scopes. To some extent I understand that scopes can be grounded, for security and, I suspect, as a leftover from the days when they had tubes and high voltages floating (pun intended) inside even when not connected to anything. What really annoy me, however, are the function generators. These are practically always low voltage, low power devices and their outputs have no need to be connected to ground, or when multichannel, to each others. It seems that a typical use for a signal generator would be to inject a signal to a DUT and check what it looks like at different points in the circuit with a scope, and it would be useful to show the input signal next to measured one. That's often not possible, and I can't imagine why. Our power supplies are typically floating, our DMMs are floating, there should be only one grounded instrument and the scope is a good candidate. Everything else should float, that would actually be safer.
I know there are signal generators with differential outputs but they fall in two camps: The very cheap ones, powered by battery or wall wart, are obviously not grounded but they are slow and very basic. The fast ones with good features that also offer floating outputs tend to be quite expensive, and that spec is often not even mentioned. They do tell you, however, never to modify the plug or use it if the ground wire is not solidly connected to earth - as their lawyers dictated - never mind that it makes it much less useful.