doc: partially revert eb20dc44

Bring back the exists gotcha, but not the references to the
`racket/exists' language.

pkgs/racket-doc/scribblings/guide/contracts/exists.scrbl

@@ -56,3 +56,4 @@ To ensure that the queue representation is abstract, we can use @racket[#:∃] i
 Now, if clients of the data structure try to use @racket[car] and @racket[cdr], they
 receive an error, rather than mucking about with the internals of the queues.
 
+See also @ctc-link["exists-gotcha"].

pkgs/racket-doc/scribblings/guide/contracts/gotchas.scrbl

@@ -108,6 +108,20 @@ Moral: if two values with contracts should interact,
        put them in separate modules with contracts at
        the module boundary or use @racket[#:freevar].
 
+@ctc-section[#:tag "exists-gotcha"]{Exists Contracts and Predicates}
+
+Much like the @racket[eq?] example above, @racket[#:∃] contracts
+can change the behavior of a program.
+
+Specifically,
+the @racket[null?] predicate (and many other predicates) return @racket[#f]
+for @racket[#:∃] contracts, and changing one of those contracts to @racket[any/c]
+means that @racket[null?] might now return @racket[#t] instead, resulting in
+arbitrarily different behavior depending on how this boolean might flow around
+in the program.
+
+Moral: Do not use predicates on @racket[#:∃] contracts.
+
 @ctc-section{Defining Recursive Contracts}
 
 When defining a self-referential contract, it is natural to use