This patch makes the following changes (according to the task description):
- Adds Type_handler::Item_func_round_fix_length_and_dec().
- Splits the code from Item_func_round::fix_length_and_dec() into new
  Item_func_round methods fix_arg_int(), fix_arg_decimal(), fix_arg_double().
- Calls the new Item_func_round methods from the relevant implementations of
  Type_handler_xxx::Item_func_round_fix_length_and_dec().
- Adds a new error message ER_ILLEGAL_PARAMETER_DATA_TYPE_FOR_OPERATION
- Makes ROUND() return the new error for GEOMETRY

Additionally:
- Inherits Item_func_round directly from Item_func_numhybrid as it
  uses nothing from Item_func_num1.
- Fixes "MDEV-12000 ROUND(expr,const_expr_returning_NULL) creates DOUBLE(0,0)".
  Now if args[1] returns NULL, the data type is set to DOUBLE with
  NOT_FIXED_DEC decimals instead of 0 decimals.

If a window function with aggregation is over the result
set of a grouping query then the argument of the aggregate
function from the window function is allowed to be an
aggregate function itself.

This patch also fixes:
 MDEV-11815 SP variables of temporal data types do not replicate correctly
Temporal values are now printed in temporal literal format, with the
SQL-standard data type prefix:
  TIME'10:20:30', DATE'2001-01-01', TIMESTAMP'2001-01-01 10:20:30'
Previously temporal values were printed using the text string notation, e.g.
 _latin1'10:20:30' COLLATE latin1_swedish_ci, hence the bug.

A GCM encrypted ciphertext must contain an authentication tag with AES_BLOCK_SIZE length, so we need to check that the length of ciphertext is at least AES_BLOCK_SIZE.

Import and adapt the changes from MySQL 5.7.

This fixes memory leaks in tests that cause InnoDB startup to fail.

buf_pool_free_instance(): Also free buf_pool->flush_rbt, which would
normally be freed when crash recovery finishes.

fil_node_close_file(), fil_space_free_low(), fil_close_all_files():
Relax some debug assertions to tolerate !srv_was_started.

innodb_shutdown(): Renamed from innobase_shutdown_for_mysql().
Changed the return type to void. Do not assume that all subsystems
were started.

que_init(), que_close(): Remove (empty functions).

srv_init(), srv_general_init(): Remove as global functions.

srv_free(): Allow srv_sys=NULL.

srv_get_active_thread_type(): Only return SRV_PURGE if purge really
is running.

srv_shutdown_all_bg_threads(): Do not reset srv_start_state. It will
be needed by innodb_shutdown().

innobase_start_or_create_for_mysql(): Always call srv_boot() so that
innodb_shutdown() can assume that it was called. Make more subsystems
dependent on SRV_START_STATE_STAT.

srv_shutdown_bg_undo_sources(): Require SRV_START_STATE_STAT.

trx_sys_close(): Do not assume purge_sys!=NULL. Do not call
buf_dblwr_free(), because the doublewrite buffer can exist while
the transaction system does not.

logs_empty_and_mark_files_at_shutdown(): Do a faster shutdown if
!srv_was_started.

recv_sys_close(): Invoke dblwr.pages.clear() which would normally
be invoked by buf_dblwr_process().

recv_recovery_from_checkpoint_start(): Always release log_sys->mutex.

row_mysql_close(): Allow the subsystem not to exist.

Display a NOTE, not a WARNING for non-default page size.

The necessary call was inadvertently commented out in a
merge of MySQL 5.7.14 to MariaDB 10.2.1
(commit fec844ac).

As noted in MDEV-11947, we should disable the InnoDB doublewrite buffer
during the tests, because when rewriting page checksums, innochecksum
would skip the pages that are in the doublewrite buffer area. Because
the doublewrite buffer is emptied on server startup and not shutdown,
we should initially start with the doublewrite buffer disabled, so that
there will be no warning messages for wrong page checksums in the
doublewrite buffer.

Also, correct the obvious typo where restart_options should have been
$restart_parameters, so that InnoDB is actually verifying that the
checksums were rewritten.

Now ROWS is a reserved key word, so cannot be used
as an alias name.

This bug happens due to a conflict in the construct window_spec.
(win_ref conflicts with the non-reserved key word ROWS).
The standard SQL-2003 says that ROWS is a reserved key word.
Made this key word reserved in our grammar and removed
the conflict.

Now, after the implementation of mdev-8646
"Re-engineer the code for post-join operations"
create_sort_index() can be called for subqueries.

Adding keywords the following keywords into the "keyword" rules:
- EXCLUDE
- UNBOUNDED
- PRECEDING
- FOLLOWING
- TIES
- OTHERS

They are non-reserved words in the SQL standard (checked in a SQL-2011 draft),
and they don't cause any conflicts in sql_yacc.yy.

Call trx_purge_state() only once, to avoid a race condition if the
value changes while the conditions are being evaluated.

Also, work around MDEV-11948 by disabling native asynchronous I/O.

Remove the debug parameter innodb_force_recovery_crash that was
introduced into MySQL 5.6 by me in WL#6494 which allowed InnoDB
to resize the redo log on startup.

Let innodb.log_file_size actually start up the server, but ensure
that the InnoDB storage engine refuses to start up in each of the
scenarios.

this fixes occasional main.cte_recursive failures

this fixes the crash in innodb.101_compatibility test

on shutdown it might happen that
1. the server starts killing THDs
2. it sets thd->killed in srv_purge_coordinator
3. srv_purge_coordinator notices that and tells srv_workers to exit
4. srv_worker will notice that and will start exiting,
    ... assert here ...
5. server sets thd->killed in worker threads

that is, it might happen that the assert is tested before
srv_worker's THD got the kill signal.

this fixes various random crashes (on this assertion) on shutdown
in tests

ninja doesn't like when a dummy target name matches a generated file name

fix the check (must be have_ not HAVE_)

also remove -O1 and -Wno-error, they aren't not needed for asan

also move *.zip files from t/ to std_data/

The temporary tables created for recursive table references
should be closed in close_thread_tables(), because they might
be used in the statements like ANALYZE WITH r AS (...) SELECT * from r
where r is defined through recursion.

that failed main.mysqlbinlog

1. wait for thd_destructor_proxy thread to start after creating it.
   this ensures that the thread is ready to receive a shutdown signal
   whenever we want to send it.
2. join it at shutdown, this guarantees that no innodb THD will exist
   after innobase_end().

this fixes crashes and memory leaks in main.mysqld_option_err
(were innodb was started and then immediately shut down).

As noted in MDEV-8841, any test that kills the server must issue
FLUSH TABLES, so that tables of crash-unsafe storage engines will
not be corrupted. Consistently issue this statement after any
call mtr.add_suppression() calls.

Also, do not invoke shutdown_server directly, but use helpers instead.

Skip test if system can't do trim operation.

InnoDB would refuse to start up if there is a mismatch on
the size of the system tablespace files. However, before this
check is conducted, the system tablespace may already have been
heavily modified.

InnoDB should perform the size check as early as possible.

recv_recovery_from_checkpoint_finish():
Move the recv_apply_hashed_log_recs() call to
innobase_start_or_create_for_mysql().

innobase_start_or_create_for_mysql(): Test the mutex functionality
before doing anything else. Use a compile_time_assert() for a
sizeof() constraint. Check the size of the system tablespace as
early as possible.

recv_scan_log_recs(): Remember if redo log apply is needed,
even if starting up in innodb_read_only mode.

recv_recovery_from_checkpoint_start_func(): Refuse
innodb_read_only startup if redo log apply is needed.

Do wait only if innodb_num_page_compressed_trim_op shows that
we have succeed to do at least few trim operations (and
that will happen on insert if possible).

Added flush tables to avoid corruption of mtr/test_suppressions
table.