The previously-working behavior without type bounds was actually somewhat odd (as explained by the added comments), so it doesn't need to work anymore.
The actual update is just to the example which uses <= instead of < in the paper. That avoids the oddity.
This also includes a small Lincons1 fix in strengthening to make the behavior consistent between octagons and polyhedra.
Closes #1373.
The previously-working behavior without type bounds was actually somewhat odd (as explained by the added comments), so it doesn't need to work anymore. The actual update is just to the example which uses
<=instead of<in the paper. That avoids the oddity.This also includes a small
Lincons1fix in strengthening to make the behavior consistent between octagons and polyhedra.