The variable gain function of the oscilloscope can be used to trim the DC common mode rejection of the probes which usually contributes the largest error. If necessary, the AC common mode rejection can be slightly adjusted with the probe compensation.
The dedicated differential probe should have a much higher common mode input voltage range for a given gain. When using the oscilloscope in add and invert mode, the maximum sensitivity is limited by the common mode voltage and not the differential voltage. Older differential probes tend to be better in this regard than newer ones or at least newer cheap ones. A 7A13 for instance can operate with a sensitivity of 100mV/div over a common mode range of +/- 1000 volts with x10 probes but an oscilloscope alone will be like 500 times worse. I suspect cheap probes leave this specification out because it makes them look bad.
I cannot really comment about the Rigols being slow in add and invert mode because none of my old DSOs slow down when operating like that.
If you are making 120 or 240 VAC differential measurements, I would first pick up a pair of inexpensive x10 or x100 probes and see how well they work before buying the more expensive differential probe. You probably should have the normal oscilloscope probes anyway and if they are good enough, you will have saved some money.