I do a tiny little bit of Sum and Difference processing (hate referring to it as Mid/Side, as that’s a mic technique and just makes things more confusing, Sum & Difference is a far more appropriate term I think), but it’s mainly volume levels (e.g. slightly drop the level of the Sum signal, maybe up to 0.5dB, for a slight stereo width enhancement).
For EQ it’s 95% regular stereo. What I do sometimes do in mastering, if needed, is the classic “monoing the bass” technique. This sometimes creates a level drop in the low end, so I might later just boost everything below 100 Hz for example, but only in the Sum signal, to compensate etc.
Other Sum and Difference techniques with EQ are only applied if there’s a problem that absolutely can’t be fixed in the mix, e.g. A) Centre Snare is too loud, so can reduce just those frequencies, just in the Sum signal, or B) Hard panned Hats are too quiet, so can boost just those frequencies, just in the Difference signal etc.
Voxengo MSED it a great freeware plugin to play with all this stuff, it can Encode, Decode, or work Inline as a Sum & Difference processor. You can set it up to encode, run it out to your hardware EQs/compressors, come back in, and Decode back to regular Stereo etc.
As an aside, I used to have lots of fun with this and my modular synth, great for experimental stuff. I would run a whole finished track through MSED in Encode mode, so it became separate Sum and Difference signals. I’d run then each of these through DIFFERENT modular synth patches. Then they’d come back into the PC and I’d use another instance of MSED to get back to regular Stereo. You’d get all sorts of very crazy stereo width/panning effects, things like the stereo width increasing as the resonant filter opened up etc. One of my fave experimental techniques, along with a Frequency Shifter in the feedback path of a Bucket Brigade Delay. 
