Also fixed a bug in row_counts detection, when partition changes.

The percent_rank function now is compatible with the cursor algorithm.
We no longer need a special implementation for it to work.

Instead of relying solely on top bound and bottom bound cursors,
now we create a list of cursors that are iterated over.

- Hook window function computation into the right location.
- Add a testcase which shows that HAVING is now checked before
  the window function computation step.

- Make EXPLAIN code use the post-join operations
- Remove Sort_and_group_tracker that was used for that purpose

man page updates: mysqlbinlog options, mysqlcheck clarification and 10.2 updates

We can call setup_partition_border_check() from
JOIN::make_aggr_tables_info(), provided that call is made after
appropriate set_items_ref_array() call.

gbh != NULL is handled in a separate branch above,
we don't get into this part of code when gbh!=NULL

Added class Window_funcs_computation, with setup() method to setup
execution, and exec() to run window function computation.

setup() is currently trivial. In the future, it is expected to optimize
the number of sorting operations and passes that are done over the temp.
table.

as it can return negative values.
The typo was introduced in the patch for MDEV-9665 in 10.2.0.

as they are not used any more.
We now use function templates from strcoll.ic instead.

Fixed calculating pointer to memory allocated for names in numeric format in unpacking procedure.

Item_func_or_sum.
Implemented method update_used_tables for class Item_findow_func.
Added the flag Item::with_window_func.
Made sure that window functions could be used only in SELECT list
and ORDER BY clause.
Added test cases that checked different illegal placements of
window functions.

- Make Item_XXX::cleanup() clean List<Cached_item> orderby_fields.
  (Items survive across PS re-executions. Cached_item don't, because
   they keep pointers to fix_field'ed items, etc)
- Move List<Cached_item> out into Group_bound_tracker.

this fixes 40 failures in mysql-test

Added --thread_cache_size=0 to some tests that was depending on not having
a thread cache.

Added --thread_cache_size=0 to slow_launch_time_func.test as otherwise
the new thread would start too fast to be counted as a slow_launch_thread.

between gbk_chinese_ci and big5_chinese_ci.

> commit e0929951
> Author: Alexander Barkov <bar@mariadb.org>
> Date:   Wed Mar 16 10:55:12 2016 +0400
>
>     MDEV-9665 Remove cs->cset->ismbchar()
>     Using a more powerfull cs->cset->charlen() instead.

        The cause of the issue is when DROP DATABASE takes
        metadata lock and is in progress through it's
        execution, a concurrently running CREATE FUNCTION checks
        for the existence of database which it succeeds and then it
        waits on the metadata lock.  Once DROP DATABASE writes to
        BINLOG and finally releases the metadata lock on schema
        object, the CREATE FUNCTION waiting on metadata lock
        gets in it's code path and succeeds and writes to binlog.