Problem:
Procedure which uses stack of views first executed without most deep view.
It fails but one view cached (as well as whole procedure).
Then simultaniusely create the second view we lack and execute the procedure.
In the beginning of procedure execution the view is not yet created so
procedure used as it was cached (cache was not invalidated).
But by the time we are trying to use most deep view it is already created.
The problem with the view is that thd->select_number (first view was not parsed) so second view will get the same number.

The fix is in keeping the thd->select_number correct even if we use cached views.
In the proposed solution (to keep it simple) counter can be bigger then should but it should not create problem because numbers are still unique and situation is very rare.

and thus reusing Used_tables_and_const_cache for Item_sum instead of
declaring the same members inside Item_sum.

and move the error reporting where it belongs

* update *.result files
* fix XtraDB for Windows (again)

* OSX (mysqlimport freeing unallocated memory)
* Windows (didn't compile MSI)
* fulltest2 (innodb crashes in --embedded --big)

Just "Master" could be understood as the master IP or hostname and thus can
cause confusion to db admins. "Master connection name" clearly states that
the log line contains connection name in the (possibly) multi-master setup.

it was never doing anything anyway

- Added --start option to mysqld which don't prints notes to log on startup
  This helps to find errors in configure options easier
- Dont write [Note] entries to log after we have abort the server
  This makes it easier to find what went wrong
- Don't by default write out Changed limits for max_open_files as this didn't really change from anything the user gave us
- Dont write warnings about not using --explicit_defaults_for_timestamp (we don't have plans do depricate the old behaviour)

Better warning from file_key_management plugin if filename is not given

Added to the pam_user_map module.

'feedback_http_proxy' system variable added to specify the
proxy server as host:port. Not a dynamic one.

- Moving Item_xxx_field declarations after Item_sum_xxx declarations,
  so Item_xxx_field constructors can be defined directly in item_sum.h
  rather than item_sum.cc. This removes some duplicate code, e.g.
  initialization of the following members at constructor time:
  name, decimals, max_length, unsigned_flag, field, maybe_null.
- Adding Item_sum_field as a common parent for Item_avg_field and
  Item_variance_field
- Deriving Item_sum_field directly from Item rather that Item_result_field,
  as Item_sum_field descendants do not need anything from Item_result_field.
- Removing hybrid infrastructure from Item_avg_field,
  adding Item_avg_field_decimal and Item_avg_field_double instead,
  as desired result type is already known at constructor time
  (not only at fix_fields time). This simplifies the code.
- Changing Item_avg_field_decimal::val_int() to call val_int_from_decimal()
  instead of doing { return (longlong) rint(val_real()); }
  This is the fix itself.

infrastructure, though in fact they always return REAL result.
Removing hybrid type artifacts.

make_cond_for_info_schema() does preserve outer fields

MDEV-8685 MariaDB fails to decode Anonymous_GTID entries
MDEV-5705 Replication testing: 5.6->10.0

- Ignoring GTID events from MySQL 5.6+ (Allows replication from MySQL 5.6+ with GTID enabled)
- Added ignorable events from MySQL 5.6
- mysqlbinlog now writes information about GTID and ignorable events.
- Added more information in error message when replication stops because of wrong information in binary log.
- Fixed wrong test when write_on_release() should flush cache.

(even when configured with --binlog-format=statement).
Before we got an error on the slave and the slave stopped if the master
was configured with --binlog-format=mixed or --binlog-format=row.