storage/oqgraph into 10.0

- since oqgraph depends on Judy, move it to a separate package (DEB)

            storage/oqgraph into 10.0

- fixed sys_vars.all_vars (added oqgraph_allow_create_integer_latch_basic.test)
- since oqgraph depends on Judy, move it to a separate package

- restore find_ticket() private status, declare friend function instead
- added metadata_lock_info to DEFAULT_SUITES

This test needs at least 320M for tokudb-max-lock-memory.
Normally tokudb-max-lock-memory is auto-sized to be 1/16th
of the available RAM size, and many our test VMs have 4G of RAM.

Do not calculate selectivity of conditions for the tables of the information schema.

The problem reported as MDEV-5444.

- include metadata_lock_info plugin into debian packages
- ignore metadata_lock_info plugin in mysqld--help test
- let test suite run with built-in metadata_lock_info plugin

The fix for bug #27937 was incomplete: it did not handle correctly the queries
containing UNION with global ORDER BY in subselects.

(better to have an address in the output than ??:0)

(PLUGIN_VAR_MEMALLOC is 0x8000 and cannot be saved in a char as such)"

https://github.com/Tokutek/ft-engine/issues/153

the system locale (on Linux) and code pages (on Windows).

modified:
  storage/connect/mysql-test/connect/t/odbc_postgresql.test
  storage/connect/user_connect.cc

modified:
  storage/connect/plgdbutl.cpp

- Update type translation to take care of type_modifiers

modified:
  storage/connect/ha_connect.cc
  storage/connect/myutil.cpp
  storage/connect/myutil.h
  storage/connect/odbconn.cpp

don't reset interrupted_query after sending the KILL signal, otherwise
the client won't know it has to stop fetching and printing the data.

added:
  storage/connect/mysql-test/connect/r/odbc_postgresql.result
  storage/connect/mysql-test/connect/t/have_odbc_postgresql.inc
  storage/connect/mysql-test/connect/t/odbc_postgresql.sql
  storage/connect/mysql-test/connect/t/odbc_postgresql.test
modified:
  storage/connect/odbconn.cpp

modified:
  storage/connect/connect.cc
  storage/connect/ha_connect.cc
  storage/connect/mysql-test/connect/r/mysql.result
  storage/connect/mysql-test/connect/r/unsigned.result
  storage/connect/mysql-test/connect/t/mysql.test
  storage/connect/mysql-test/connect/t/unsigned.test

does not exist.

with the same names present in multiple schemas

The "TABNAME" option now supports qualified table names,
to connect to tables residing in a particular schema and catalog.

Qualified table names have the following format:

  [[CatalogName.]SchemaName.]TableName

Qualified table names can be used:

1. In "normal" tables:

CREATE TABLE t1 ENGINE=CONNECT TABLE_TYPE=ODBC
CONNECTION='DSN=ConnectEng;UID=mtr;PWD=mtr'
TABNAME='schema1.t1';

2. In catalog tables (CATFUNC=Tables  and CATFUNC=Columns)

CREATE TABLE t1 ENGINE=CONNECT TABLE_TYPE=ODBC
CONNECTION='DSN=postgresql;UID=user;PWD=password'
TABNAME='schema1.t1';

Note, the % and _ wildcards are supported in
the schema name and the table name parts:

CREATE TABLE t1 ENGINE=CONNECT TABLE_TYPE=ODBC
CONNECTION='DSN=postgresql;UID=user;PWD=password'
TABNAME='%.t1';

CREATE TABLE t1 ENGINE=CONNECT TABLE_TYPE=ODBC
CONNECTION='DSN=postgresql;UID=user;PWD=password'
TABNAME='schema1.%';

Removed unused_tables, find LRU object by timestamp instead.

- tc_acquire_table and tc_release_table do not access
  TABLE_SHARE::tdc.used_tables anymore
- in tc_acquire_table(): release LOCK_tdc after we relase LOCK_open
  (saves a few CPU cycles in critical section)
- in tc_release_table(): if we reached table cache threshold, evict
  to-be-released table without moving it to unused_tables. unused_tables
  must be empty at this point.

fix the element deleting logic for roles_mappings_hash

use generic restore_ptr_backup() approach

MDEV-4403 Attempting to use cassandra storage engine causes "service 'my_snprintf_service' interface version mismatch"

When a DSO is loaded we rewrite service pointers to point to the actual service structures.
But when a DSO is unloaded, we have to restore their original values, in case this DSO
wasn't removed from memory on dlclose() and is later loaded again.