Movegen speedup via king attack info

[ddobbelaere]

Various movegen optimizations centered around the introduction of precomputed king attack info (attacking lines, pinned pieces, double check detection) that is used to speedup several parts of legal move detection (for all moves).

Unconditional speedup for random backend up to 6% (tested with LTO builds on Linux Fedora 29 64-bit, gcc 8.2.1).

The following measurements all share the command line lc0 benchmark --backend=random --smart-pruning-factor=0 --nodes=1000000 --nncache=1000000. A 95% confidence interval for the average of the last reported nps value over 25 runs is given for the starting position, the opening position https://lichess.org/editor/rn2kb1r/pp3ppp/2p1pnb1/8/3P1N1q/4B1N1/PPPQ1PPP/R3KB1R_b_KQkq_-_5_9 and the endgame position https://lichess.org/editor/8/2R2P2/1p4K1/6R1/1r6/5rP1/1k6/2q5_w_-_-_0_55 .

The measurements are performed on an Intel Core i5-5300U CPU @ 2.30GHz with 4 cores and L1d cache: 32K, L1i cache: 32K, L2 cache: 256K, L3 cache: 3072K.

For 1 search thread:

position	master 6a639b6	this PR	speedup
start	208222 ± 448	215247 ± 726	3.4%
opening	161321 ± 216	168776 ± 335	4.6%
endgame	176833 ± 245	186576 ± 283	5.5%

For 2 search threads:

position	master 6a639b6	this PR	speedup
start	271137 ± 830	272221 ± 882	0.4%
opening	224128 ± 422	227968 ± 551	1.7%
endgame	238111 ± 511	243892 ± 758	2.4%

[ddobbelaere]

This PR is ready for review.

[mooskagh]

On my machine chessboard_test with uncommented tests:
master:

[ RUN      ] ChessBoard.MoveGenStartingPos
[       OK ] ChessBoard.MoveGenStartingPos (6676 ms)
[ RUN      ] ChessBoard.MoveGenKiwipete
[       OK ] ChessBoard.MoveGenKiwipete (11287 ms)
[ RUN      ] ChessBoard.MoveGenPosition3
[       OK ] ChessBoard.MoveGenPosition3 (868 ms)
[ RUN      ] ChessBoard.MoveGenPosition4
[       OK ] ChessBoard.MoveGenPosition4 (1052 ms)
[ RUN      ] ChessBoard.MoveGenPosition5
[       OK ] ChessBoard.MoveGenPosition5 (5088 ms)
[ RUN      ] ChessBoard.MoveGenPosition6
[       OK ] ChessBoard.MoveGenPosition6 (202 ms)

With this PR

[ RUN      ] ChessBoard.MoveGenStartingPos
[       OK ] ChessBoard.MoveGenStartingPos (5931 ms)
[ RUN      ] ChessBoard.MoveGenKiwipete
[       OK ] ChessBoard.MoveGenKiwipete (9873 ms)
[ RUN      ] ChessBoard.MoveGenPosition3
[       OK ] ChessBoard.MoveGenPosition3 (739 ms)
[ RUN      ] ChessBoard.MoveGenPosition4
[       OK ] ChessBoard.MoveGenPosition4 (892 ms)
[ RUN      ] ChessBoard.MoveGenPosition5
[       OK ] ChessBoard.MoveGenPosition5 (4413 ms)
[ RUN      ] ChessBoard.MoveGenPosition6
[       OK ] ChessBoard.MoveGenPosition6 (184 ms)

Looks good!

(for the reference, uncommented lines:

--- a/src/chess/board_test.cc
+++ b/src/chess/board_test.cc
@@ -94,7 +94,7 @@ int Perft(const ChessBoard& board, int max_depth, bool dump = false,
 }
 }  // namespace
 
-/* TEST(ChessBoard, MoveGenStartingPos) {
+TEST(ChessBoard, MoveGenStartingPos) {
   ChessBoard board;
   board.SetFromFen(ChessBoard::kStartposFen);
 
@@ -105,7 +105,7 @@ int Perft(const ChessBoard& board, int max_depth, bool dump = false,
   EXPECT_EQ(Perft(board, 4), 197281);
   EXPECT_EQ(Perft(board, 5), 4865609);
   EXPECT_EQ(Perft(board, 6), 119060324);
-} */
+}
 
 TEST(ChessBoard, MoveGenKiwipete) {
   ChessBoard board;
@@ -116,7 +116,7 @@ TEST(ChessBoard, MoveGenKiwipete) {
   EXPECT_EQ(Perft(board, 2), 2039);
   EXPECT_EQ(Perft(board, 3), 97862);
   EXPECT_EQ(Perft(board, 4), 4085603);
-//  EXPECT_EQ(Perft(board, 5), 193690690);
+  EXPECT_EQ(Perft(board, 5), 193690690);
 }
 
 TEST(ChessBoard, MoveGenPosition3) {
@@ -151,7 +151,7 @@ TEST(ChessBoard, MoveGenPosition5) {
   EXPECT_EQ(Perft(board, 2), 1486);
   EXPECT_EQ(Perft(board, 3), 62379);
   EXPECT_EQ(Perft(board, 4), 2103487);
-//  EXPECT_EQ(Perft(board, 5), 89941194);
+  EXPECT_EQ(Perft(board, 5), 89941194);
 }
 
 TEST(ChessBoard, MoveGenPosition6) {

)