My Rainbow Serpent _034 had no trouble finding just the first move a6, only considers a6, but thinks it is a draw only. Giving it some more moves does not help, until after 7. Bg3 of the above official line:peter wrote:Very tricky!Vinvin wrote: 5)
Zugzwang too : 1B1k4/3p4/1Kp5/PpP2p2/1PqP1P2/8/1P6/8 w - - bm a6;
[pgn]
[Event "?"]
[Site "?"]
[White "w"]
[Black "b"]
[Result "*"]
[SetUp "1"]
[FEN "1B1k4/3p4/1Kp5/PpP2p2/1PqP1P2/8/1P6/8 w - - 0 1"]
1. a6 Kc8 2. a7 Qa2 3. Bd6 Qa4 4. Be7 Qa2 5. Bh4 Qa4 6. b3 Qa3 7. Be1 Qa2
8. Bd2 Qa1 9. Bc1 Qa2 10. Bb2 *
[/pgn]
Found it in YACPDB
http://www.yacpdb.org/#search/MUIxazQzc ... MS8xLzA=/1
with a slightly different line of solution from O.Bláthy, in which the longer way with one more black pawn- move is shown.
1. a6 Kc8 2. a7 Qa2 3. Bd6 Qa4 4. Be7 Qa2 5. Bh4 Qa4
6. b3 Qa3 7. Bg3 Qa2 8. Bf2 Qa1 9. Be1 Qa2 10. Bd2 Qa1
11. Bc1 d5 12. Be3
In this one (or in your line above by 9...d5 instead of Qa2), you can fool even engines with switched off or very little nullmove by default like komodo-NM-off or Sting for quite a while yet.
Even if the eval already remains a winning one, much to much shuffling still then.
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "1B1k4/3p4/1Kp5/PpP2p2/1PqP1P2/8/1P6/8 w - -"]
1. a6 Kc8 2. a7 Qa2 3. Bd6 Qa4 4. Be7 Qa2 5. Bh4 Qa4 6. b3
Qa3 7. Bg3 *
Even then , it lost the solution, until finding it back at depth 33
33 1:59 +12.43 7...Qa2 8.Bf2 Qa3 9.Be3 Qa2 10.Bd2 Qa1 11.Bc1 d5 12.Bd2 Qa2 13.Be1 Qa1 14.Bh4 Qa3 15.Bg3 Qa2 16.Bf2 Qa3 17.Be3 Qa2 18.Bd2 Kd8 19.Kb7 Qxd2 20.a8Q+ Ke7 (1.316.326.062) 11052
32 1:59 0.00 7...Qa1 8.Be1 (1.316.326.062) 11052
32 1:59 +39.61 7...d6 8.cxd6 Qa1 9.Bf2 Kd7 10.Kb7 Qh1 11.a8Q c5+ 12.Kb8 Qxa8+ 13.Kxa8 cxb4 14.Kb7 Kxd6 15.Kb6 Kd5 16.Kxb5 Ke4 17.Kc6 Kxf4 18.d5 Kg5 19.d6 Kh5 20.d7 f4 (1.316.326.062) 11052
32 1:59 +39.61 7...d5 8.cxd6 Qa1 9.Bf2 Kd7 10.Kb7 Qh1 11.a8Q c5+ 12.Kb8 Qxa8+ 13.Kxa8 cxb4 14.Kb7 Kxd6 15.Kb6 Kd5 16.Kxb5 Ke4 17.Kc6 Kxf4 18.d5 Kg5 19.d6 Kh5 20.d7 f4 (1.316.326.062) 11052
Stockfish Master (well, with a small change, reversal of a simplification doubled pawns handling) sees absolutely still nothing here, 4 best moves, single thread mode.
[d]2k5/P2p4/1Kp5/1pP2p2/1P1P1P2/qP4B1/8/8 b - -
Engine: Sf20160727doublePawn (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
59 54:26 0.00 7...Qa2 8.Bh4 Qa1 9.Bf2 Qa3 10.Bg3 (11.678.603.250) 3575
59 54:26 0.00 7...Qa1 8.Bf2 Qa3 9.Bg3 (11.678.603.250) 3575
58 54:26 +M34 7...d6 8.cxd6 Qa1 9.d7+ Kxd7 10.Kb7 Qh1
11.a8Q c5+ 12.Ka7 Qa1+ 13.Kb8 Qxd4
14.Qb7+ Ke6 15.Qc8+ Kd6 16.Qc7+ Kd5
17.Bf2 Qh8+ 18.Kb7 Ke4 19.bxc5 Kf3
20.c6 Qh1 (11.678.603.250) 3575
58 54:26 +M33 7...d5 8.cxd6 Qa1 9.d7+ Kxd7 10.Kb7 Qh1
11.a8Q c5+ 12.Ka7 Qa1+ 13.Kb8 Qxd4
14.Qb7+ Ke6 15.Qc8+ Kd6 16.Qc7+ Kd5
17.Bf2 Qh8+ 18.Kb7 Ke4 19.bxc5 Kf3
20.c6 Qh1 (11.678.603.250) 3575
best move: Qa3-a2 time: 55:16.906 min n/s: 3.575.100 nodes: 11.861.047.763
Well, both did not very good but with a bit less pruning, the bench goes up a little but not like disabling nullmove, a new version was born. It still has very (very) high Nullmove R but now sometimes skips the whole process. One line change. Maybe I finally got a hint of the never published Fruit nullmove trick

Engine: Rainbow Serpent 20160626_034 MP
Based on 014 and 028 -bmi2 TDM-GCC
===========================
Total time (ms) : 4470
Nodes searched : 9611954
Nodes/second : 2150325
New version:
Engine: Rainbow Serpent 20160626_035 MP
Based on 014 and 028 -bmi2 TDM-GCC
===========================
Total time (ms) : 5003
Nodes searched : 10953268
Nodes/second : 2189339
The new version exhibits "some" search instability but that is a small price to pay compared to not finding anything until move 7
1B1k4/3p4/1Kp5/PpP2p2/1PqP1P2/8/1P6/8 w - -
Engine: Rainbow Serpent 20160626_035 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
30/18 0:01 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (17.871.246) 10811
31/18 0:02 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (27.778.538) 10923
32/18 0:02 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (29.901.153) 10980
33/18 0:03 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (37.730.662) 11003
34/18 0:05 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (62.758.554) 11060
35/18 0:09 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (103.804.762) 11250
36/18 0:12 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (138.186.822) 11203
37/18 0:12 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (139.028.813) 11203
38/18 0:15 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (172.957.973) 11129
39/18 0:27 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (307.965.831) 11360
40/18 0:32 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (369.823.309) 11342
41/18 1:01 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (701.427.294) 11381
42/18 1:37 0.00 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bb8 (1.101.194.902) 11274
43/18 1:50 +0.07++ 1.a6 (1.242.740.828) 11251
43/18 2:09 +0.15++ 1.a6 (1.470.087.914) 11319
43/18 2:10 +0.27++ 1.a6 (1.472.804.866) 11319
43/37 2:23 +0.43++ 1.a6 (1.620.328.944) 11315
43/54 2:35 +0.65++ 1.a6 (1.758.362.781) 11319
43/59 2:55 +0.94++ 1.a6 (1.984.300.666) 11299
43/59 3:20 +1.33++ 1.a6 (2.263.628.042) 11265
43/59 3:47 +1.84++ 1.a6 (2.556.672.086) 11240
43/59 4:32 +2.48++ 1.a6 (3.047.594.925) 11196
43/59 5:29 +3.31++ 1.a6 (3.693.247.248) 11194
43/59 6:38 +4.36++ 1.a6 (4.456.389.021) 11173
43/66 8:07 +5.69++ 1.a6 (5.441.312.105) 11155
43/67 9:01 +4.59-- 1.a6 Kc8 (6.035.944.023) 11140
43/67 9:46 +5.98++ 1.a6 (6.530.036.927) 11131
43/67 13:09 +7.22 1.a6 Kc8 2.a7 Qa2 3.Bc7 Qa1 4.b3 Qa3
5.Be5 Qa2 6.Bf6 Qa3 7.Bh4 Qa1 8.Be1 Qa2
9.Bd2 Qa1 10.Bc1 d5 11.Be3 Qa3
12.Bf2 Qa2 13.Bg3 Qa3 14.Bh4 (8.721.364.836) 11046
44/67 17:48 +7.15-- 1.a6 Kc8 (11.776.568.619) 11023
44/67 18:22 +7.07-- 1.a6 Kc8 (12.160.390.218) 11028
44/67 19:09 +6.95-- 1.a6 Kc8 (12.686.990.280) 11035
44/67 19:34 +6.79-- 1.a6 Kc8 (12.955.637.961) 11031
44/67 19:34 +6.49-- 1.a6 Kc8 (12.958.874.973) 11031
44/67 19:34 +6.17-- 1.a6 Kc8 (12.960.631.938) 11031
44/67 19:34 +5.63-- 1.a6 Kc8 (12.961.112.355) 11031
44/67 19:35 +5.23-- 1.a6 Kc8 (12.962.023.565) 11031
44/67 19:49 +5.68++ 1.a6 (13.119.875.330) 11034
44/67 19:51 +5.13-- 1.a6 Kc8 (13.146.541.625) 11034
44/67 19:58 +5.82++ 1.a6 (13.220.663.430) 11034
44/67 20:00 +4.95-- 1.a6 Kc8 (13.246.335.609) 11034
44/67 20:28 +6.05++ 1.a6 (13.547.307.905) 11026
44/67 20:41 +4.66-- 1.a6 Kc8 (13.684.026.116) 11025
44/67 21:55 +2.00-- {

44/67 22:12 +4.20++ {

44/85 22:58 +4.15 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bc7 Qa2
5.Bd6 Qa3 6.Be7 Qa2 7.Bh4 Qa1 8.Be1 d5
9.Bh4 Qa3 10.Bf6 Qa2 11.Bg5 Qa1
12.Bd8 Kxd8 13.Kb7 Qxd4 14.a8Q+ (15.168.652.527) 11005
.
.
.
45/85 67:42 +12.79 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Be5 Qa2
5.Bf6 Qa3 6.Bh4 Qa2 7.Bf2 Qa3 8.Be3 Qa2
9.Bd2 Kd8 10.Bc3 Kc8 11.Bb2 Kd8
12.Kb7 Qxb2 13.a8Q+ Ke7 14.Kc7 (43.561.801.212) 10723
.
.
.
46/85 89:13 +13.50 1.a6 Kc8 2.a7 Qa2 3.b3 Qa3 4.Bd6 Qa1
5.Be5 Qa2 6.Bf6 Qa3 7.Be7 Qa2 8.Bh4 Qa3
9.Bg3 Qa2 10.Bf2 Kd8 11.Be1 Kc8
12.Bd2 Kd8 13.Bc3 Ke7 14.Kb7 (57.156.455.839) 10676
.
.
.
48/92 121:24 +13.76 1.a6 Kc8 2.a7 Qa2 3.Bd6 Qa4 4.b3 Qa2
5.Bb8 Qa1 6.Be5 Qa2 7.Bf6 Qa3 8.Bh4 Qa1
9.Be1 Qa2 10.Bd2 Qa1 11.Bc1 d5
12.Be3 Qa3 13.Bf2 Qa2 14.Bh4 (77.352.783.309) 10618
.
.
.
49/92 302:44 +16.39 1.a6 Kc8 2.a7 Qa2 3.Bc7 Qa1 4.Be5 Qa2
5.Bf6 Qa1 6.b3 Qa2 7.Be7 Qa1 8.Bg5 Qa2
9.Bf6 Qa3 10.Bh4 Qa2 11.Bf2 Qa3
12.Be3 d6 13.Bd2 Qa1 14.Bc3 (192.546.735.928) 10600
FH, FL, FH etc.
Maybe it was search luck that it now found at least the proper score, not a good line probably.