Bugfix in surround.c

[nando]

The appended patch fixes a conceptual bug I introduced a long time ago in the surround code. Under circumstances, dragons which should have been cataloged as "not surrounded" were actually still considered as "weakly surrounded".

The breakage is neutral, 1 PASS and 1 FAIL.

trevor:220      PASS E8 [E8]
strategy4:190   FAIL E7 [D13]

I just can say that the failure is definitely not a problem with the bugfix itself.

Other than that, these are the counters I'm getting with my Linux box:

Total nodes: 1691312768 3342139 12618757
Total time: 7832.19 (7840.57)
Total uncertainty: 46.88
1 PASS
1 FAIL

-- nando

[nando]

Bugfix in compute_surroundings()

[gunnar]

Clearly a bugfix, indeed. It might be simpler to do that board loop in 1D instead of 2D though.

[nando]

Patch revised according to Gunnar's comment

[nando]

Certainly, patch revised accordingly.