…l number of entries per ply, not used in QSearch

… also uses the TT.

Score of MinimalChess 0.5.5 vs Bit-Genie_5: 1297 - 1787 - 916  [0.439] 4000 with option.Hash=10 -each tc=5+0.5
(prev version with 2 buckets and no Qsearch also scored 0.439... so not a definitive improvement)

…inversely proportional to the depth. So leaf nodes have search-depth more slots than root nodes. There's no depth = -1 anymore for Qsearch nodes. Instead QSearch now also uses the TT with depth 0. On the ECM test positions this seemed to perform better then the previous implementation. In actual games no improvement could be shown but maybe the time controls are too small (less than 1s per move) for this implementation's theoretical strength to compensate the overhead of the more complex implementation over simpler systems like for example a 2-bucket or even always replace TT.

…of MinimalChess 0.5.6b vs Bit-Genie_5: 1353 - 1788 - 859 [0.446] 4000 which is the best result any TT variation achieved so far.

…cket solution. This commit simplifies the previous TT implementation to 2 buckets. The replacement scheme is still based on a combination of age and how many positions of the same ply are already stored. If age is equal it favors the position that has less "sibling" in the TT.

…h and age. This version tested strong and is fairly simple and is the one going into the master branch.