MDEV-16188 Use in-memory PK filters built from range index scans

This patch contains a full implementation of the optimization
that allows to use in-memory rowid / primary filters built for range  
conditions over indexes. In many cases usage of such filters reduce  
the number of disk seeks spent for fetching table rows.

In this implementation the choice of what possible filter to be applied  
(if any) is made purely on cost-based considerations.

This implementation re-achitectured the partial implementation of
the feature pushed by Galina Shalygina in the commit
8d5a11122c32f4d9eb87536886c6e893377bdd07.

Besides this patch contains a better implementation of the generic  
handler function handler::multi_range_read_info_const() that
takes into account gaps between ranges when calculating the cost of
range index scans. It also contains some corrections of the
implementation of the handler function records_in_range() for MyISAM.

This patch supports the feature for InnoDB and MyISAM.

1 files changed, 3 insertions, 1 deletions

diff --git a/mysql-test/main/subselect.test b/mysql-test/main/subselect.test
index d7c266cafb0..dd80b789516 100644
--- a/mysql-test/main/subselect.test
+++ b/mysql-test/main/subselect.test
@@ -2026,6 +2026,8 @@ insert into t1 values (10);
 
 create table t2 (a int primary key, b varchar(32), c int, unique key b(c, b));
 insert into t2(a, c, b) values (1,10,'359'), (2,10,'35988'), (3,10,'35989');
+insert into t2(a, c, b) values (4,10,'360'), (5,10,'35998'), (6,10,'35999');
+analyze table t1;
 
 explain SELECT sql_no_cache t1.a, r.a, r.b FROM t1 LEFT JOIN t2 r
   ON r.a = (SELECT t2.a FROM t2 WHERE t2.c = t1.a AND t2.b <= '359899'
@@ -3878,7 +3880,7 @@ CREATE TABLE t3 (
   KEY b_2 (b)
 );
 
-INSERT INTO t3 VALUES (1,1,1), (2,32,1);
+INSERT INTO t3 VALUES (1,1,1), (2,32,1), (3,33,1), (4,34,2);
 
 explain
 SELECT t1.a, (SELECT 1 FROM t2 WHERE t2.b=t3.c AND t2.c=t1.a ORDER BY t2.d LIMIT 1) AS incorrect FROM t1, t3 WHERE t3.b=t1.a;