do not return negative mate too

[jbienvenue]

fix #1121

STC non-regr
Elo | 2.16 +- 2.27 (95%)
SPRT | 8.0+0.08s Threads=1 Hash=16MB
LLR | 3.06 (-2.25, 2.89) [-2.75, 0.25]
Games | N: 21846 W: 5616 L: 5480 D: 10750
Penta | [73, 2286, 6071, 2418, 75]
https://recklesschess.space/test/15624/

LTC non-regr
Elo | 2.16 +- 2.28 (95%)
SPRT | 40.0+0.40s Threads=1 Hash=64MB
LLR | 2.80 (-2.25, 2.89) [-2.75, 0.25]
Games | N: 18960 W: 4679 L: 4561 D: 9720
Penta | [8, 1920, 5516, 2018, 18]
https://recklesschess.space/test/15625/

Bench: 2569707

[jbienvenue]

for information, matetrack on main :

Found mate #14 (better) for FEN "8/4p3/8/8/8/5p1p/5K2/N6k w - -"  with bm #16.

Using /home/jojo/myreck/recklessmain on matetrack.epd with --nodes 1000000
Engine ID:     Reckless 0.10.0-dev-1298b3ca
Total FENs:    6554
Found mates:   3711
Best mates:    2120

Parsing the engine's full UCI output, the following issues were detected:
Better mates:                1   (from 1 FENs)

matetrack on this patch :

Using /home/jojo/myreck/reckless on matetrack.epd with --nodes 1000000
Engine ID:     Reckless 0.10.0-dev-7cbe1c66
Total FENs:    6554
Found mates:   3727
Best mates:    2117

[robertnurnberg]

Given that there was a bug, it might be prudent to run the (old) script test_engine.sh from vondele/matetrack#36, for both single- and multi-threaded analysis.

[jbienvenue]

no problem with this program with threads 1, 2, 4, 8, 16