Similarly to 5d4edba925, it seems
that sometimes MoltenVK returns 0 to timestamp queries. It
doesn't happen deterministically and it might depend on the
hardware (I have seen differences between the M2 I used some
time ago and the M3 Max I have now).
My main motivation to this is avoiding generating a lot of useless
log lines from other executors when I'm interested in just one of
them, but I can imagine this also somewhat improving efficiency.
This test is currently miscompiling on SM4 because
copy_propagation_invalidate_variable_from_deref_recurse() is not always
invalidating the right components.
Basically, separate lower_casts_to_int() into the lowering of the CAST
and the lowering of the TRUNC, so that TRUNCs that are not part of a
cast are lowered as well.
Adds new flags --sm-min and --sm-max. They each take a shader model
identifier, with the same syntax as in the test harness. If either is
present, then it will only run tests within the (inclusive) range.
Omitting one allows anything as the min/max.
While fxc allows full expressions inside the angle brackets, we don't parse that
yet as it'd be quite a mess to properly do so with yacc, and I'm not aware of any
game doing so in their shaders.
