bug#36139: [PATCH] Make better use of the switch op in cond forms

[Stefan Monnier]

> Allow switch generation with a mixture of eq, eql, equal, memq, memql and 
> member.

AFAIK all of those give the same result (in practice) as using
equal/member: I'm having a hard time imagining an eq/eql/memq/memql test
against a constant which behaves differently from equal/member except
for those that can simply always return nil (e.g. (eq x "toto") can
always return nil since there's no way the caller of this code can make
sure x is really the same string object as the "toto" generated by the
compiler).

So I think we should "standardize" on equal/member and mostly disregard
the eq/eql/equal difference (except maybe for emitting a warning when
comparing against something where `equal` doesn't give the same result).


        Stefan