Small semeai revisions

[nando]

A simple idea: if we aren't going to use shape_offensive_moves, maybe shouldn't we look for them in the first place.

The breakage is neutral, but not as neutral as one would expect. Due to the nature of the change, a variation in the number of OWL nodes appears at first as not very logical. But we know by experience that the persistent reading cache may at times induce such effects. I'm confident it's the case here, but if someone insists on getting proof of that, I'll try to find time to dig deeper.

Total nodes: 1697336351(-0.6%) 3336337(+eps) 12311921(-1.4%)

The other change is a typo-fix in the second call to get_next_move_from_list(). The owl parameter was wrong. This has no consequences, since cuts aren't used currently within the semai code.

[nando]

Minor semeai code revisions

[gunnar]

Replying to nando:

The breakage is neutral, but not as neutral as one would expect. Due to the nature of the change, a variation in the number of OWL nodes appears at first as not very logical. But we know by experience that the persistent reading cache may at times induce such effects. I'm confident it's the case here, but if someone insists on getting proof of that, I'll try to find time to dig deeper.

Don't bother digging into this. The patch looks fine, and gives a nice reduction of nodes.