It is not simply the case that the main line in the oscilloscope probe carries a signal, and that the ground line is just something to catch noise.
Both lines carry the signal.
We need two wires out of a device, because inside that device, there are two points between which a voltage is developed which is understood as some signal.
To bring that signal to another device, we have to convey both points, and so we need two conductors.
We call one of these lines "ground" or "return" or "common". But electrons do not know that; they do not read schematics. Electrons do not know that this line is intended to carry signal, and that other one catches noise. It doesn't work that way.
If you sever the ground, the whole arrangement of the oscilloscope and the DUT will still exhibit some kind of "signal". The scope continues to measure the potential difference between the probe tip and the ground clip. However the ground clip is no longer the correct voltage point against which the signal is properly defined.
Any two points in space have some potential difference between them (possibly zero), regardless of whether current can easily flow between them.
The oscilloscope scope is a high impedance input which requires very little current to flow. It measures pure voltage, or almost. Without a ground in place, the scope then still measures the potential between two points in space: the disconnected ground and the point in the DUT being probed. The potential between those two points includes stuff other than the intended signal, like noise induced by stray electric fields.