I have my 2 ground loop stories:
1) Hot and Neutral and Ground rotated on my ancient HP bench I got used when I joined the R&D team. Didn't know it and didn't seem to do any harm until I connected a BNC reference cable between two benches. It wasn't good.
2) In China, in a high-tech lab, my customer was complaining that the PNA reading was jumping up and down, very noisy; I kept changing the sweep rate until it was about 200 msec full screen, guess what, 10 cycles exactly (50 Hz, China, remember). I said "maybe a ground loop"; they said " not possible, look, nothing else connected." This is why you have to go there in person. I peak my head around the back of the wafer prober (where the PNA was) and there was a 10 MHz reference BNC cable daisy-chained from one station to the next. I disconnect that and the noise disappears. They had been dealing with this for something like 3 months and were ready to send the unit back.
Here's the thing, much (most?) equipment has bypass capacitors from Hot to Chassis-ground and from Neutral to Chassis-ground, I guess for conducted emissions, and they are of course regulated so the value is set that the max current flow in the ground can't be more than a milliamp or something (don't hold me to that) and so if you have a power system that is not grounded, the chassis really will float at 1/2 line value.
In fact (make a long story longer), there was an issue where it seemed HP 8753 VNAs source amplifiers were blowing up, especially overseas (like Japan). Turns out it is very common (at that time, 1980's) to not have grounded mains in Japan. So the chassis was floating at 55 V, and if someone put a test cable on, that some how was earth ground on the center pin, you would see 55 V across the RF connector
to chassis and blow the main amplifier (we did a re-design to put a clamp on that, but really it was a bit of a challenge to do that and not affect the 3 GHz performance). When I suggested to our environmental test group that we add a test to check if grounding the RF center pin of an instrument when the chassis is ungrounded could be added to the test suite their response is "An ungrounded chassis is a fault condition, we don't test instruments when they are powered in a faulty condition". Eventually, they agreed to add it to the "strife-test" which is where we intentionally test-to-failure (the impressive strife test is putting a 50 kG, $400k PNA on a shake table and vibrating it until something breaks, always exciting that).