The following left in semi-improved state to keep patch size reasonable:
- Field operator new: left thd_alloc(current_thd)
- Sql_alloc operator new: left thd_alloc(thd_get_current_thd())
- Item_args constructors: left thd_alloc(thd)
- Item_func_interval::fix_length_and_dec(): no THD arg, have to call current_thd
- Item_func_dyncol_exists::val_int(): same
- Item_dyncol_get::val_str(): same
- Item_dyncol_get::val_int(): same
- Item_dyncol_get::val_real(): same
- Item_dyncol_get::val_decimal(): same
- Item_singlerow_subselect::fix_length_and_dec(): same

Part2: moving a few other fields from Column_definition to Create_field.
sizeof(sp_variable) is now 200 bytes (vs 248 bytes in 10.1)

- MDEV-8093 sql_yacc.yy: add %type create_field for field_spec and column_def

and partially:

- MDEV-8095 Split Create_field

and don't set Create_field->sql_type to MYSQL_TYPE_VIRTUAL
temporarily only to change it again few lines later.

see commit 81e4ce5e

A prerequisite change for:
- MDEV-8093 sql_yacc.yy: add %type create_field for field_spec and column_def
- MDEV-8094 sql_yacc.yy: get rid of the rules "opt_if_not_exists_table_element"
            and "opt_if_exists_table_element"
- MDEV-8095 Split Create_field

(that is MY_CHECK_C_COMPILER_FLAG, MY_CHECK_CXX_COMPILER_FLAG,
MY_CHECK_AND_SET_COMPILER_FLAG)

This is to have a clear "internal" name prefix and to avoid
mixing them with hand-named symbols

A preparatory task for:
  MDEV-4912 Add a plugin to field types (column types)

wsrep.cmake should be included before cpack_rpm.cmake as it defines
WITH_WSREP, required by cpack_rpm.cmake to conditionally include galera
helper tools in server package's requires list.

--with-plugin-name=xxxx
--with-plugins= now uses =AUTO instead of =1

1. define connect_EXPORTS, this causes the engine to use MariaDB
   versions for timestamp<->struct tm conversion instead of
   TZ-dependent libc versions.
2. remove check_access() that was removed once, but re-appeared
   during a complex merge.
3. disable a totally broken test
4. update test results
5. skip odbc_firebird test when no firebird DSN is available

Apply MySQL fix for bug#76727:
https://github.com/mysql/mysql-server/commit/69d4e72c

  Bug#20901025: SLAVE ASSERTION IN UNPACK_ROW WITH ROLLBACK TO
  SAVEPOINT IN ERROR HANDLER

feedback was bzero-ing thd->status_vars.memory_used,
but server was asserting the correct value of it