Problem: when character_set_connection=utf8,
mixing SPACE() with a non-Unicode column (e.g. for concat)
produced "illegal mix of collations" error.
Fix: Item_string() corresponding to space character
is now created using "ASCII" repertoire. Previously
it was incorrectly created using "UNICODE" repertoure, which
didn't allow to convert results of SPACE() to a non-Unicode
character set.


mysql-test/include/ctype_common.inc:
  - Adding test for bug#27580
  - Restoring previous values of character_set_client and character_set_results,
  because ctype_common.inc now changes them when doing "set names utf8"
  in the test for bug#27580
mysql-test/r/ctype_big5.result:
  Adding test
mysql-test/r/ctype_cp1250_ch.result:
  Adding test
mysql-test/r/ctype_euckr.result:
  Adding test
mysql-test/r/ctype_gb2312.result:
  Adding test
mysql-test/r/ctype_gbk.result:
  Adding test
mysql-test/r/ctype_uca.result:
  Adding test
mysql-test/r/ctype_ucs.result:
  Adding test
mysql-test/t/ctype_cp1250_ch.test:
  Adding test
mysql-test/t/ctype_ucs.test:
  Adding test
sql/item_create.cc:
  Item for SQL function SPACE() is now created with ASCII repertoire,
  to allow automatic conversion from UTF8 to column's character
  set e.g. for CONCAT().