added:

  @ mysql-test/include/ctype_heap.inc
  @ mysql-test/r/ctype_utf16le.result
  @ mysql-test/t/ctype_utf16le.test
  Adding rests

modified:
  @ cmake/character_sets.cmake
  Adding utf16le into the list of the possible default character sets
  and into the list of complex character sets.

  @ include/m_ctype.h
  Adding prototype for global variables

  @ mysql-test/r/ctype_utf16.result
  Conversion from binary to utf16 did not work well. 
  Warning did not work well.
  Recording new results

  @ mysql-test/r/ctype_utf32.result
  Warning did not work well.
  Recording new results

  @ mysys/charset-def.c
  Adding initialization for utf16le collations

  @ sql/item.cc
  Warning did not work well for non-ASCII based character sets.
  Adding ErrConvString.

  @ sql/item_func.cc
  Error message about bad truncated number in CAST(string AS SIGNED)
  did not work well (a missing part of "WL#751 Error message construction").

  @ sql/sql_string.cc
  Conversion from binary to utf16 and utf16le did not work well.
  0x61 was incorrectly extended to 0x00000061 instead of 0x0061.

  @ sql/sys_vars.cc
  "SET collation_connection=utf16_general_ci" followed by
  "SET NAMES 'latin1'" did not work well,
  because get_charset_by_csname('latin1') was called with utf16 representation
  of 'latin1' string. Fixing val_str() to val_str_ascii() to always call
  get_charset_by_csname() with 8-bit representation.
  Fixing the similar problem for get_charset_by_name().

  @ strings/ctype-ucs2.c
  - Changing my_strtoll10_mb2 from accessing memory directly using
  s[0] and s[1] to mb_wc() method, to reuse the same code for utf16le.
  - Adding useful macro for surrogate high/low first/last values.
  - Adding macros MY_UTF16_WC2 and MY_UTF16_WC4, to make some utf16
    routines more readable.
  - Fixing my_caseup_utf16, my_hash_sort_utf16, my_ismbchar_utf16,
    my_strnncoll_utf16_bin, my_hash_sort_utf16_bin to use cs->cset->mb_wc
    instead of dirrect call to my_utf16_uni, to reuse the code.
  - Fixing my_hash_sort_utf16 and my_hash_sort_utf16_bin to use lengthsp()
    instead of direct memory access to reuse the code for utf16le.
  - Adding low level utf16le functions my_utf16le_uni, my_uni_utf16le,
    my_lengthsp_utf16le. These are the only three new functions.
    All other functions reuse utf16/ucs2 code.
  - Adding utf16le structures.