bug#20351: isearch-lax-whitespace should be customizable variable

[Drew Adams <drew.adams@oracle.com>]

> > The technical term is "equivalence classes".
> 
> I don't think so.  This term would apply equally well to case-fold-
> search.  We want something more specific.

Right.  But isn't that the problem?  Case folding is a subset
of character folding, where the equivalence class is defined
by ignoring letter case.

This new thing is a catch-all, no?  It means (apparently)
all character equivalences except for case folds.

We should foresee the possibility (and I hope the reality)
of multiple other character foldings, besides case folding.

And yes, each should have a name that suggests the kind of
equivalence that is involved.

What can we say about the kind of equivalence that is involved
in this catch-all?  How can we be more specific than, say,
"char folding other than case" - e.g., non-case char-folding?