*: speed up the operation of "admin check table"

[zimulala]

What problem does this PR solve?

Reduce the execution time of this operation of "admin check table".

What is changed and how it works?

compare the result of "select count() from table" and "select count() from table use index(idx)"

• table count == index count
  We will check the records with the IndexLookUpReader method of all indexes.

• table count > index count
  We will use the current index's checkTableRecord to check the records.

• table count < index count
  We will use the current index's IndexLookUpReader to check the records.

Using 3 TiKV(8 Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz), 1 PD, 1 TiDB , the servers of TiKV and TiDB are not on a server.

CREATE TABLE `x` (
  `a` int(11) NOT NULL,
  `b` varchar(255) DEFAULT NULL,
  `c` blob DEFAULT NULL,
  `d` time DEFAULT NULL,
  `e` double DEFAULT NULL,
  `f` text DEFAULT NULL,
  `g` bit(10) DEFAULT NULL,
  PRIMARY KEY (`a`),
  KEY `d` (`d`),
  KEY `a_2` (`a`,`b`),
  KEY `c` (`c`(10),`d`),
  KEY `e` (`e`,`f`(100)),
  KEY `a_3` (`a`,`b`,`c`(2048)),
  KEY `d_2` (`d`,`e`,`f`(1024),`g`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin |

before vs after

mysql> select count(*) from x use index(d);
+----------+
| count(*) |
+----------+
| 10000000 |
+----------+
1 row in set (2.30 sec)

before this PR:
mysql> admin check table x;
More than 3 hours

after this PR:
mysql> admin check table x;
Query OK, 0 rows affected (45.52 sec)

Executing this operation with old code takes too long, so the following only tests the new code

mysql> select count(*) from x use index(d);
+-----------+
| count(*)  |
+-----------+
| 100000000 |
+—————+
Query OK, 0 rows affected (4 min 27.76 sec)

CREATE TABLE `lineitem` (
  `L_ORDERKEY` int(11) NOT NULL,
  `L_PARTKEY` int(11) NOT NULL,
  `L_SUPPKEY` int(11) NOT NULL,
  `L_LINENUMBER` int(11) NOT NULL,
  `L_QUANTITY` decimal(15,2) NOT NULL,
  `L_EXTENDEDPRICE` decimal(15,2) NOT NULL,
  `L_DISCOUNT` decimal(15,2) NOT NULL,
  `L_TAX` decimal(15,2) NOT NULL,
  `L_RETURNFLAG` char(1) NOT NULL,
  `L_LINESTATUS` char(1) NOT NULL,
  `L_SHIPDATE` date NOT NULL,
  `L_COMMITDATE` date NOT NULL,
  `L_RECEIPTDATE` date NOT NULL,
  `L_SHIPINSTRUCT` char(25) NOT NULL,
  `L_SHIPMODE` char(10) NOT NULL,
  `L_COMMENT` varchar(44) NOT NULL,
  PRIMARY KEY (`L_ORDERKEY`,`L_LINENUMBER`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_bin 

mysql> select count(*) from TPCH10.lineitem;
+----------+
| count(*) |
+----------+
| 59986052 |
+----------+
1 row in set (4.69 sec)
tidb> admin check table TPCH10.lineitem;
Query OK, 0 rows affected (1 min 16.92 sec)

mysql> select count(*) from TPCH50.lineitem;
+-----------+
| count(*)  |
+-----------+
| 300005811 |
tidb> admin check table TPCH50.lineitem;
Query OK, 0 rows affected (5 min 50.57 sec)

Check List

Tests

• Unit test

Side effects

• Possible performance regression

Related changes

• Need to cherry-pick to the release branch
• Need to update the documentation
• Need to be included in the release note

---

This change is 

[winkyao]

Really impressive.

[winkyao]

Please fix ci.

[zimulala]

/run-all-tests

[zimulala]

PTAL @winkyao @jackysp

[zimulala]

/run-unit-test

[jackysp]

OK, great work! I need more time to review it.

[zimulala]

/run-unit-test

[zimulala]

PTAL @winkyao @jackysp @crazycs520 @XuHuaiyu

[winkyao]

@zimulala Please address comments

[zimulala]

PTAL @winkyao @AilinKid

[AilinKid]

This pr is too long, needing time to deliberate. By now it's LGTM

[zimulala]

/run-all-tests

[winkyao]

LGTM

[sre-bot]

/run-all-tests

[sre-bot]

@zimulala merge failed.

[winkyao]

/run-unit-test

[sre-bot]

cherry pick to release-3.0 failed