[ranges] Consistent style for negated fold expressions in constraints

[CaseyCarter]

I'm reasonably certain that LWG has forwarded papers for C++23 that use the syntax (for a type pack class... Ts):

  requires (!(meow && ... && Ts))

and other papers that use:

  requires (!meow || ... || !Ts)

These forms are functionally equivalent. I think we should pick one syntax for such negated folds and use it consistently.

[CaseyCarter]

Tim Song suggested in LWG:

I personally prefer !(X && ...) though since those are usually paired with another overload that's (Y &&...), and I find it easier to compare the two without De Morgan'ing the former. YMMV

which seems like a good argument to prefer the conjunctive form over the disjunctive.

[frederick-vs-ja]

I think they become less equivalent since C++26 due to P2963R3...