[m-rev.] for review: edit_seq for --warn-inconsistent-pred-order

[Zoltan Somogyi]

For review by anyone.

This should fix one test case failure for C# (and probably Java), because
io.call_system_code does not handle redirection.

A question: is it ok for the warning to list long sequences of unchanged lines,
or should I postprocess the output to elide them.

A related question: does anyone insist on the current organization of io.m,
where foreign_procs are clustered by language? I would like to reorder its
contents to put all the foreign_procs for a predicate next to each other.
That should make it easier to catch problems like this where the different
foreign_procs for the same predicate do slightly different things.

Zoltan.