part(x) should give warning

part(x) returns x.

This is of course perfectly consistent (the null case),
but it is certainly an error if actually input this way.

True, it might be useful in the case apply(part,cons
(expression,specifier)), where specifier is an argument,
possibly the empty list, but using apply this way is
somewhat unclean. I would prefer that there be an
explicit form of part where the specifier is a list. The
problem with that solution, of course, is that we'd then
need corresponding versionf of substpart, inpart,
substinpart. Yecch.

For now, I would recommend giving a warning message
for this case. It would be nice if there could, however,
be only one such warning per interaction. I don't think
we do anything like that right now.