[m-rev.] for review: type_spec_constrained_preds

[Zoltan Somogyi]

This diff includes fixes for the problems revealed by the tryout,
and has been cleaned up to be ready for review. For review by Julien,
since he is already familiar with the new pragma.

This diff does not include user-facing documentation of the pragma.
My plan is that we should use the new pragma ourselves for a week or so,
replacing plain type_spec pragmas in existing code (Julien's csv and json libs,
the Mercury library and compiler) to find any problems that we haven't
found so far, before we announce its existence. I also want to look into
making type_spec pragmas composable. I have an approach that
I think will work, and if it does, then I would prefer to document
just *that* version of the new pragma, instead of both the non-composable
version and then the composable version.

Zoltan.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Log.tscp
Type: application/octet-stream
Size: 7375 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20240206/38822b6d/attachment-0002.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: DIFF.tscp
Type: application/octet-stream
Size: 191709 bytes
Desc: not available
URL: <http://lists.mercurylang.org/archives/reviews/attachments/20240206/38822b6d/attachment-0003.obj>