[m-rev.] for review: pragma consider_used

[Julien Fischer]

Hi Zoltan,

On Fri, 11 Dec 2015, Zoltan Somogyi wrote:

> The only things that needs review are the concept and the name;
> the implementation is straightforward.
>
> I will add public documentation once we have a consensus on
> the name.

The concept and name are fine by me.

> As the log message says, I have tested the diff on stratify.m.
> After two or three weeks, when this diff has been installed
> on all our development systems, I intend to add these pragmas
> to the compiler itself, so separate test cases aren't needed.

Not quite; presumably the pragmas cannot refer to non-existent
predicates, so there should be a test case add to tests/invalid for
that.

Julien.