Fix some problems in the TC_LOG_MMAP commit processing, which could
lead to assertions in some cases.

Problems are mostly reproducible in MariaDB 10.0 with asynchroneous
commit checkpoints, but most of the problems were present in earlier
versions also.

Adjust unstable test case.

Properly drop all unused keys.
Patch by Igor Babaev.

followup fixes for MySQL Bug #13889741: HANDLE_FATAL_SIGNAL IN _DB_ENTER_ | HANDLE_FATAL_SIGNAL IN STRNLEN

Take into account that length of strings passed down to this function can be up to FN_REFLEN+1 bytes. including terminating zero.
The overwrite was caused by incomplete fix to MySQL Bug # 44834

If triggers are used for an insert/update/delete statement than the values of
all virtual columns must be computed as any of them may be used by the triggers.

add a test case for MySQL Bug #13889741: HANDLE_FATAL_SIGNAL IN _DB_ENTER_ | HANDLE_FATAL_SIGNAL IN STRNLEN

  
The problem is that memory alocated by copy_andor_structure() well be freed,
but if level of SELECT_LEX it will be excluded (in case of merge derived tables and view)
then sl->where/having will not be updated here but still can be accessed (so it will be access to freed memory).

(patch by Sanja)

Made information_schema_all_engines stable by adding "sorted_result".

If, when executing a query with ORDER BY col LIMIT n, the optimizer chose
an index-merge scan to access the table containing col while there existed
an index defined over col then optimizer did not consider the possibility
of using an alternative range scan by this index to avoid filesort. This
could cause a performance degradation if the optimizer flag index_merge was
set up to 'on'.

mysql-test/r/partition.result:
  Added test case
mysql-test/t/partition.test:
  Added test case
sql/ha_partition.cc:
  Removed printing of not initialized variable
storage/maria/ha_maria.cc:
  Don't copy variables that are not initialized

storage/maria/ma_test2.c:
  Problem was that rnd() generated bigger value than allocated array