[CM] debuggin ligeti

[Tobias Kunze Briseño]

 hm.  my diffs were for clisp.  don't know what the deal is with
 cmucl.  but watch out: there is more wrong with the code than
 just the compile.  as i said, flow is different from my original
 (which is why some vars don't get initialized properly) and most
 importantly, when i ran the 2.3.4 version today, it turned out 
 twice as fast (which could be a problem within cm-2.3.4's timing
 code, or within ligeti.cm, i don't know).

 i think it may need a rewrite from the original 1.3 sources, but
 rick is superbusy and so am i...