Multiple optimizations (speed and memory usage improvement)

[facontidavide]

15% speed improvement when using the default heuristics.

[facontidavide]

It is actually faster now

Old benchmark

BM_AStar_Big          129443995 ns    129443311 ns           32
BM_AStar_Smooth_1000   91320707 ns     91319632 ns           46
BM_AStar_Small          5938943 ns      5938910 ns          706

New benchmark

BM_AStar_Big           94683926 ns     94681340 ns           73
BM_AStar_Smooth_1000   65355006 ns     65354110 ns          107
BM_AStar_Small          4174486 ns      4174392 ns         1676

[facontidavide]

By the way, we are also using 20% mess RAM in this branch.

[chataign]

I didn't check all the changes in detail but it looks good and is faster. I merged a change from master to only clear the cells that were visited, I found that it makes a big difference also. I created tag v1 with the previous version before the merge. Thanks for all the work!!