unless it's a test of SP or PS

it's already checked before the update loop (see check_fields()),
and this condition cannot suddenly become false in the middle
of the statement

followup to be81b00c
Fix updates and deletes too

changed to use DBUG keywords instead,
so that the code is compiled and tested

added tests.

This reverts commit 07b1a774.

and the system_versioning_transaction_registry variable.
The user enables transaction registry by specifying BIGINT for
row_start/row_end columns.

check mysql.transaction_registry structure on the first open,
not on startup. Avoid warnings unless transaction_registry
is actually used.

rpl.rpl_events sometimes fails in release mode after be81b00c

This could be prevented with MSAN

MDEV-14871 Server crashes in fill_record / fill_record_n_invoke_before_triggers upon inserting into versioned table with trigger

Merge branch '10.3' into trunk

* show SYSTEM VERSIONED in INFORMATION_SCHEMA.TABLES
* show ROW START/ROW END columns in INFORMATION_SCHEMA.COLUMNS

must behave as any other generated column: an error in the strict mode,
a warning otherwise.

sys_trx_start -> row_start
sys_trx_end -> row_end

use system time, not @@timestamp

rename to test_versioning
set maturity to experimental

This bug affected tables where the PRIMARY KEY contains variable-length
columns, and ROW_FORMAT is COMPACT or DYNAMIC.

rec_init_offsets_comp_ordinary(): Do not short-cut the parsing
of the record header for records that contain explicit values
for instantly added columns.

rec_copy_prefix_to_buf(): Copy more header for records that
contain explicit values for instantly added columns.

row_purge_reset_trx_id(): Display a DBUG message about resetting the
DB_TRX_ID.

same change as for PSI_TABLE_CALL

it's correct, they preserve their documented names

remove unused error messages
reword ER_VERS_SYS_FIELD_NOT_HIDDEN->ER_VERS_SYS_FIELD_EXISTS

Many related changes.

Note that AS OF condition must always be pushed down to physical tables,
it cannot be applied to a derived or a view. Thus:

* no versioning for internal temporary tables, they can never store
  historical data.
* remove special versioning code from mysql_derived_prepare and
  remove ER_VERS_DERIVED_PROHIBITED - derived can have no historical
  data and cannot be prohibited for system versioning related reasons.
* do not expand select list for derived/views with sys vers fields,
  derived/views can never have historical data.
* remove special invisiblity rules for sys vers fields, they are no
  longer needed after the previous change
* remove system_versioning_hide, it lost the meaning after the
  previous change.
* remove ER_VERS_SYSTEM_TIME_CLASH, it's no "clash", the inner
  AS OF clause always wins.
* non-versioned fields in a historical query

reword the warning text, downgrade to note, don't
replace values with NULLs

CTE don't work, report MDEV-14820

compiler warning (mix of bool and enum in ?:)

in JOIN::inject_best_splitting_cond

The value of SplM_opt_info::last_plan should be set to NULL
before any search for a splitting plan for a splittable
materialized table.

MDEV-14821 Assertion `!is_set() || (m_status == DA_OK_BULK && is_bulk_op())` failed in Diagnostics_area::set_ok_status

In case libaio is not found, and required,
remove variables HAVE_LIBAIO_H and HAVE_LIBAIO from cache, so that cmake
rerun after installation of libaio would succeed.