my_atomic_add32_explicit on windows
Fixed that server_audit.c also works if one compiles with safemalloc
Fixed compiler warnings

This is needed because of the new code where THD creation was moved to the new thread

server has started (assert in sys_var::where get_sys_var_value_origin())

Creating a CONNECT object on client connect and pass this to the working thread which creates the THD.
Split LOCK_thread_count to different mutexes
Added LOCK_thread_start to syncronize threads
Moved most usage of LOCK_thread_count to dedicated functions
Use next_thread_id() instead of thread_id++

Other things:
- Thread id now starts from 1 instead of 2
- Added cast for thread_id as thread id is now of type my_thread_id
- Made THD->host const (To ensure it's not changed)
- Removed some DBUG_PRINT() about entering/exiting mutex as these was already logged by mutex code
- Fixed that aborted_connects and connection_errors_internal are counted in all cases
- Don't take locks for current_linfo when we set it (not needed as it was 0 before)

(Benchmarked, about 30 % faster and 50 % smaller than original)

Conflicts:
	VERSION
	cmake/plugin.cmake
	config.h.cmake
	configure.cmake
	plugin/server_audit/server_audit.c
	sql/sql_yacc.yy

Revert the patch for MDEV-9504.
It causes test failures, attempt to fix these causes more failures. The
source of all this is that the code in test_if_skip_sort_order() has
a peculiar way of treating select_limit parameter:
Correct value is computed when the query plan is changed. In other cases,
we use an approximation that ignores the presence of GROUP BY clause,
or JOINs, or both.

A patch that fixes all of the above would be too big to do in 10.1

The memory loss could happen in Aria with encrypted tables when Aria failed to encrypt a block.
In normal usage this should never happen.

New capability flags space.
Removed old progress flag, added new one.

Conflicts:
	configure.cmake

create_partition_index_description() had wrong logic to calculate
length of the key value buffer that is used by the range optimizer.

For some reason it used MAX(partitioning_columns_len,
subpartitioning_columns_len) while it should use SUM of these values.

- Legacy code would set JOIN_TAB::limit only for EXPLAIN queries (this
  variable is only used when producing EXPLAIN output)

- ANALYZE/SHOW EXPLAIN need to produce EXPLAIN output for non-EXPLAIN
  queries, too, so we should always set JOIN_TAB::limit.

Conflicts:
	config.h.cmake
	configure.cmake

Conflicts:
	configure.cmake

The check for UCONTEXT in cmake was somehow become broken, disabling the
fallback to ucontext. This caused the non-blocking client API to not be
available for non-x86 platforms, on which no hand-crafted assembler
implementation of my_context is available.

DBUG_ASSERT() added in the patch for MDEV-9181 did not take into account
special circumstances for the prepared statement EXECUTE. Fixig the assert.
Also, extending and fixing comments made during MDEV-9181.

The test restarts the server, it should not be executed for embedded

This is to add compatibility with the similar MySQL option.

Don't assert if decrypt or encrypt fails if my_assert_on_error is not set.
Added failed file name if encryption/decryption fails.

Wrapping args[0] and args[2] into an Item_cache for aggregate functions.

        The SQL command 'PREPARE' was broken - should be take into
        account.

        Keep the embedded-server version valid.

        SHOW CREATE PROCEDURE/FUNCTION fixed.

        SHOW BINARY LOGS fixed.

        SHOW MASTER STATUS fixed.

        SHOW SLAVE STATUS fixed.

        Fix for SHOW GRANTS statement.

  Fix for SHOW CREATE DATABASE.

MDEV-8615: Assertion `m_cpp_buf <= begin_ptr && begin_ptr <= m_cpp_buf + m_buf_length' failed in Lex_input_stream::body_utf8_start

Nothing should be done before any keyword recognized.

This reverts commit 2ee149be4e9319208dfcfb77712aa1ef805f6ba8.