[Bug-openmcl] two one-liners for loop

[Gary Byers]

In case anyone's counting, the current failure total's down to ~75
(depending on what you mean by "current".)

- Several pathname bugs,
- a few LOOP failures
- several SUBTYPEP corner cases
- several cases where ANSI CL and the MOP differ
- (RATIONALIZE single-float-negative-epsilon) is off by a bit
- several tests where NIL is used as an array type specifier that are not
  suppressed by the :nil-vectors-are-strings note
- MACROLET (and DEFMACRO, for that matter) don't behave correctly when
  a destructuring pattern is used for &WHOLE.
- probably a few other things that are easily fixable but lost in the
  noise.

I think that when Bryan started this a few weeks back he said that the
number of failures was > 700, so this is definite improvement ...