One of the subtle things about Grip Factor that is difficult to think about and illustrate is that it (a) reduces wheel weight as you lose grip, and (b) losing the rear end is a grip loss case, but (c) when the backend does step out the front wheels will already tend to maintain their direction (but feel like they're rotating in relation to the car) and this feels like wheel weight loss because the steering rack is following the path of least resistance..
See, it's hard to describe. The physics is already in place but Grip Factor < 1.0 adds some force to accentuate the feeling.
When you add Grip Factor < 1.0 you're adding *additional* wheel weight loss above and beyond the natural 'turning' of the front wheels when the backend steps out. The *additional* wheel weight loss can make it feel unrealistic for some wheels or scenarios. That's my take.
I understand if belt/gear wheels need this, but I agree with Jul's, we should start with 1.0 and decrease in very small amounts just the amount that is needed, and then you will be closer to realistic than the default GTR2 value for that param.
On a related note, in my ffb mod, for next version (v6) I'm actually setting Grip Factor=1.0 to tune out "awkward wheel behaviour during lateral slides". If Grip Factor is not 1.0 then when you have a lateral car slide the wheel doesn't feel right. It feels like it's trying to tell you something that isn't happening. I have a feeling that people who submitted PLRs previously won't even notice the difference because we've tuned the other more apparent params first.