increase two positions' mate score from 23

[robertnurnberg]

This PR proposes to raise the bm value for two positions from 23 to 34 and 36, respectively.

The positions are very similar, and have been discussed in detail in #52.

The general view is that the puzzle setters have made some mistake, somewhere.

For the first position (8/p7/qb5p/p7/p6p/7p/2P1P2k/2Q2K2 w - -), I have done a local analysis with Huntsman to give:

info depth 97 seldepth 68 multipv 1 score mate 34 nodes 19789014004696 nps 25275350 hashfull 1000 tbhits 0 time 782937268 pv c1f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 c2c4 h2g3 c6f3 g3h2 f3f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 e2e3 b6e3 c6a6 h2g3 f1e2 e3b6 c4c5 g3f4 c5b6 a7b6 e2d3 h3h2 a6a8 f4f5 a8e4 f5f6 e4h4 f6f5 h4h5 f5f4 h5h2 f4f5 h2h5 f5e6 d3e4 e6d6 h5d5 d6c7 d5e6 a4a3 e4d5 c7b7 e6d7 b7a6 d7c8 a6a7 d5c6 a3a2 c8b7

For the second position (8/p7/qb5p/p6p/p7/7p/2P1P2k/2Q2K2 w - -), @Yery noticed in #52 (comment) that white can force this sequence of moves: c1f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 c2c4. From the resulting position (8/p7/qbQ4p/p6p/p1P5/7p/4P2k/5K2 b - c3 0 7) I have done a local analysis with Stockfish to give:

info depth 117 seldepth 63 multipv 1 score mate -29 nodes 15607966517328 nps 33119950 hashfull 1000 tbhits 0 time 471255732 pv h2g3 c6f3 g3h2 f3f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 e2e3 b6e3 c6a6 h2g3 a6g6 e3g5 g6d3 g3f4 d3h3 g5d8 c4c5 d8e7 h3h5 e7g5 f1e2 a4a3 h5f3 f4e5 f3a3 e5d5 e2d3 g5h4 a3a2 d5c5 a2c4 c5d6 c4h4 d6e5 h4d4 e5f5 d4d5 f5f6 d3e4 a5a4 d5c6 f6f7 e4f5 a4a3 c6c7 f7g8 f5f6 a3a2 c7g7

and so in total a M36 PV as follows:

c1f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 c2c4 h2g3 c6f3 g3h2 f3f4 h2h1 f4e4 h1h2 e4e5 h2h1 e5d5 h1h2 d5d6 h2h1 d6c6 h1h2 e2e3 b6e3 c6a6 h2g3 a6g6 e3g5 g6d3 g3f4 d3h3 g5d8 c4c5 d8e7 h3h5 e7g5 f1e2 a4a3 h5f3 f4e5 f3a3 e5d5 e2d3 g5h4 a3a2 d5c5 a2c4 c5d6 c4h4 d6e5 h4d4 e5f5 d4d5 f5f6 d3e4 a5a4 d5c6 f6f7 e4f5 a4a3 c6c7 f7g8 f5f6 a3a2 c7g7

Both of my local analyses confirm results also found by @Yery.

We cannot be 100 percent sure that these are indeed the shortest possible mates. But I would suggest to merge this now, and leave possible improvements for later.