MDEV-6066: Merge new defaults from 5.6 and 5.7 (defaults changed, QC can be stopped with no-zero size)

--autoset- command line prefix added

* Start_encryption_log_event
* --encrypt-binlog command line option

based on google patches.

Introduce Log_event_writer() that encapsulates
writing data to an IO_CACHE with automatic checksum calculation.

Now all events properly checksum themselves as needed.

Use Log_event_writer in MYSQL_BIN_LOG::write_cache() instead
of copy-pasting its logic all over.

Later Log_event_writer will also do encryption.

now when my_vsnprintf() supports %llu for a few years already.

commit f1abd015
Author: Andrei Elkin <aelkin@mysql.com>
Date:   Thu Nov 12 17:10:19 2009 +0200

    Bug #47210   first execution of "start slave until" stops too early

introduce DBUG_DUMP_EVENT_BUF,
remove few unused DBUG_EXECUTE_IF's
simplify few DBUG_PRINT's
remove few redundant #ifndef DBUG_OFF's

* replace binlog positions where they are calculated
  (instead of --disable_query_log)
* don't replace binlog positions when they're hard-coded anyway

1. use include/show_binlog_events.inc instead of SHOW BINLOG EVENTS
2. use include/show_relaylog_eventc.inc too
3. in all other places where a number might appear in the result
   file, include binlog_start_pos.inc, calculate the position
   like pos=`select $binlog_start_pos + 100`; and use
   replace_result $pos <pos>

and fix SHOW RELAYLOG 'src_name' EVENTS to respect src_name

* fix unireg.h includes
* use enum_binlog_checksum_alg for binlog checksum variables,
  not uint8

* reformat long lines
* remove useless 'const'
* remove unneeded do { ... } while(false} wrapper
* remove unneeded method

There are three Log_event::read_log_event() methods:
1. read the event image from IO_CACHE into String
2. create Log_event from the in-memory event image
3. read the event image from IO_CACHE and create Log_event

The 3rd was reading event image into memory and invoking the 2nd to
create Log_event. Now the 3rd also uses the 1st to read the event image
from IO_CACHE into memory, instead of duplicating its functionality.

remove my_crc_dbug_check (gdb can do it itself).
use 0 instead of my_checkum(0, 0, 0) - just as 10.0 does now.

Instead of encrypt(src, dst, key, iv) that encrypts all
data in one go, now we have encrypt_init(key,iv),
encrypt_update(src,dst), and encrypt_finish(dst).

This also causes collateral changes in the internal my_crypt.cc
encryption functions and in the encryption service.

There are wrappers to provide the old all-at-once encryption
functionality. But binlog events are often written piecewise,
they'll need the new api.

Rename reassociate to reset and create an inverse method release.
Method names are chosen to match std::unique_ptr methods.

make repeated cmake runs less verbose:
* remove few not very useful MESSAGE's
* only run pkg_check_modules() if there's no cached result
* only print QQGraph messages on the first run

Fix all cmake tests (including plugin) to use
MY_CHECK_AND_SET_COMPILER_FLAG. And fix that function
to be compatible with cmake 3.0. This way flag checks
are correctly cached (even in cmake 3.0) and properly reused.

It caused the following main.innodb_load_xa to fail.

referenced_by_foreign_key2(), needed for InnoDB to compile,
was taken from 10.0-galera

If a transaction T1 needs to wait for a transaction T2, T2's commit will
skip the normal binlog_commit_wait_usec delay, in order not to needlessly
stall throughput.

This works by checking if T2 is already ready to commit. If so, it is woken
up. If not, we set a flag in T2 so that when it gets ready to commit, it
will do so immediately.

But there was a potential race due to insufficient locking, if T2 gets ready
to commit just at the point where T1 does the check. If the race hits, the
wakeup (and early commit) of T2 might be lost.

The race is only theoretical (from code inspection, no known test case), but
seems best to fix it anyway, by properly locking LOCK_prepare_ordered around
the check.

MDEV-8725: Assertion `!(thd->rgi_slave && thd-> rgi_slave->did_mark_start_commit)' failed in ha_rollback_trans

The assertion is there to catch cases where we rollback while
mark_start_commit() is active. This can allow following event groups
to be replicated too early, causing conflicts.

But in this case, we have an _explicit_ ROLLBACK event in the binlog,
which should not assert.

We fix this by delaying the mark_start_commit() in the explicit
ROLLBACK case. It seems safest to delay this in ROLLBACK case anyway,
and there should be no reason to try to optimise this corner case.

and vcol_supported_sql_funcs_innodb. Moving the test for using CHARSET(),
COLLATION(), COERCIBILITY() in virtual column from
vcol_supported_sql_funcs_xxx to vcol_blocked_sql_funcs_xxx,
as these functions are not supported in virtual columns any longer.
Discussed with Sanja on IRC.