MDEV-29805 Attempt to insert into system versioning columns on old server may make slave data diverge

MDEV-29732 mysqlbinlog produces syntactically incorrect output with system_versioning_insert_history

don't set vers_write=false if one vers column was used explicitly,
instead do vers_update_fields() for columns that do not have explicit
value. So, if row_start has the value and row_end not, row_end will
get max by default.

* clarify the help text for --system-versioning-insert-history
* move the vers_write=false check from Item_field::fix_fields()
  next to other vers field checks in find_field_in_table()
* move row_start validation from handler::write_row() next to
  vers_update_fields()
* make secure_timestamp check to happen in one place only,
  extract it into a function is_set_timestamp_vorbidden().
* overwriting vers fields is an error, just like setting @@timestamp
* don't run vers_insert_history() for every row

1. system_versioning_insert_history session variable allows
pseudocolumns ROW_START and ROW_END be specified in INSERT,
INSERT..SELECT and LOAD DATA.

2. Cleaned up select_insert::send_data() from setting vers_write as
this parameter is now set on TABLE initialization.

4. Replication of system_versioning_insert_history via option_bits in
OPTIONS_WRITTEN_TO_BIN_LOG.

they should be set once, not for every row

ROW START/END fields shold have no default value

fix quoting

* Add query optimizer timer to ANALYZE FORMAT=JSON

* Adapt tests and results

* Change logic to always close the writer after printing query blocks

This adds a header row in txt files exported by mariadb-dump --tab

Reviewed by: Daniel Black and Dan Lenski

trx_sys_t::undo_log_nonempty: Set to true if there are undo logs
to rollback and purge.

The algorithm for re-creating the undo tablespace when
trx_sys_t::undo_log_nonempty is disabled:

1) trx_sys_t::reset_page(): Reset the TRX_SYS page and assign all
rollback segment slots from 1..127 to FIL_NULL

2) Free the rollback segment header page of system tablespace
for the slots 1..127

3) Update the binlog and WSREP information in system tablespace
rollback segment header
Step (1), (2) and Step (3) should happen atomically within a
single mini-transaction.

4) srv_undo_delete_old_tablespaces(): Delete the old undo tablespaces
present in the undo log directory

5) Make checkpoint to get rid of old undo log tablespaces redo logs

6) Assign new start space id for the undo log tablespaces

7) Re-create the specified undo log tablespaces. InnoDB uses same
mtr for this one and step (6)

8) Make checkpoint again, so that server or mariabackup
can read the undo log tablespace page0 before applying
the redo logs

srv_undo_tablespaces_reinit(): Recreate the undo log tablespaces.
It does reset trx_sys page, delete the old undo tablespaces,
update the binlog offset, write set replication checkpoint
in system rollback segment page

trx_rseg_update_binlog_offset(): Added 2 new parameters to pass
binlog file name and binlog offset

trx_rseg_array_init(): Return error if the rollback segment
slot points to non-existent tablespace

srv_undo_tablespaces_init(): Added new parameter mtr
to initialize all undo tablespaces

trx_assign_rseg_low(): Allow the transaction to use the rollback
segment slots(1..127) even if InnoDB failed to change to the
requested innodb_undo_tablespaces=0

srv_start(): Override the user specified value of
innodb_undo_tablespaces variable with already existing actual
undo tablespaces

wf_incremental_process(): Detects whether TRX_SYS page has been
modified since last backup. If it is then incremental backup
fails and throws the information about taking full backup again

xb_assign_undo_space_start(): Removed the function. Because
undo001 has first undo space id value in page0

Added test case to test the scenario during startup and mariabackup
incremental process too.

Reviewed-by : Marko Mäkelä
Tested-by : Matthias Leich

Fix test cases to remove hardcoded expected number for 'handler_next_read'
status variable. Apparently, number of rows in I_S.routines depends on
whether or not 'sys' schema is loaded, and this depends on whether or not
perfschema is compiled in.

- Cosmetic fixes

Reviewer:Brandon Nesterenko <brandon.nesterenko@mariadb.com>

- Add `replicate_rewrite_db` status variable, that may accept comma
separated key-value pairs.
- Note that option `OPT_REPLICATE_REWRITE_DB` already existed in `mysqld.h`
  from this commit 23d8586d

Reviewer:Brandon Nesterenko <brandon.nesterenko@mariadb.com>

The mysqld_multi script template used @datadir@ as default log
destination, this is not the MariaDB datadir in this context though
but rather the -- typically write-only -- /share dir.

The correct placeholder to use here is @localstatedir@ which gets
replaced with the actual MariaDB datadir
Co-authored-by: Hartmut Holzgraefe <hartmut@php.net>