Fixes to tests and their results, to account for differences between InnoDB 1.0.4 and the old builtin.
All committed result differences have either been verified by me or copied from Oracle's provided
results (storage/innodb_plugin/mysql-test/*.result, storage/innodb_plugin/mysql-test/patches).

mysql-test/r/information_schema.result:
  queries changed a bit
mysql-test/r/information_schema_db.result:
  queries changed a bit
mysql-test/r/innodb-autoinc.result:
  importing piece from storage/innodb_plugin/mysql-test/innodb-autoinc.result
mysql-test/r/innodb.result:
  result close to storage/innodb_plugin/innodb.result, except 4095 pages instead of 8191, which makes
  sense as Summit runs with a buffer pool of 64M, whereas the mentioned result file was made with
  a buffer pool of 128M.
mysql-test/r/mysqlshow.result:
  InnoDB I_S tables have arrived
mysql-test/suite/funcs_1/r/is_columns_is.result:
  queries changed a bit
mysql-test/suite/funcs_1/r/is_columns_is_embedded.result:
  queries changed a bit
mysql-test/suite/funcs_1/r/is_tables_is.result:
  queries changed a bit
mysql-test/suite/funcs_1/t/is_columns_is.test:
  making I_S queries ignore InnoDB I_S tables
mysql-test/suite/funcs_1/t/is_columns_is_embedded.test:
  making I_S queries ignore InnoDB I_S tables
mysql-test/suite/funcs_1/t/is_tables_is.test:
  making I_S queries ignore InnoDB I_S tables
mysql-test/suite/innodb/r/innodb-zip.result:
  result update
mysql-test/suite/innodb/t/innodb-zip.test:
  439, as size of prefix key, throws error with certain system zlib (ubuntu
  "intrepid") but not with zlib bundled with MySQL, because zlib's
  compressBound() are different (and used by InnoDB's page_zip_empty_size()).
mysql-test/suite/sys_vars/r/innodb_file_per_table_basic.result:
  result update
mysql-test/suite/sys_vars/r/innodb_lock_wait_timeout_basic.result:
  result update
mysql-test/suite/sys_vars/r/innodb_sync_spin_loops_basic_32.result:
  result update (default value is 30 in the plugin, 20 in the builtin)
mysql-test/suite/sys_vars/r/innodb_sync_spin_loops_basic_64.result:
  result update (default value is 30 in the plugin, 20 in the builtin)
mysql-test/suite/sys_vars/r/table_definition_cache_basic.result:
  result update (default value is 400 in Summit)
mysql-test/suite/sys_vars/t/innodb_file_per_table_basic.test:
  variable is writable in the plugin (patch from Oracle)
mysql-test/suite/sys_vars/t/innodb_lock_wait_timeout_basic.test:
  variable is per-session in the plugin (patch from Oracle)
mysql-test/t/information_schema.test:
  making I_S queries ignore InnoDB I_S tables
mysql-test/t/information_schema_db.test:
  making I_S queries ignore InnoDB I_S tables
mysql-test/t/innodb-autoinc.test:
  importing piece from storage/innodb_plugin/mysql-test/innodb-autoinc.test
mysql-test/t/innodb.test:
  importing most of storage/innodb_plugin/t/innodb.test. Most replace_result were not needed (no path printed), some where (for --embedded).
mysql-test/t/mysqlshow.test:
  output of test now depends on InnoDB or not InnoDB. As there is no way to make mysqlshow
  produce a single output in those two cases (no way to make it exclude InnoDB I_S
  tables), let the test depend on InnoDB, it isn't a very selective condition, and the
  test is simple enough.
storage/innobase/CMakeLists.txt:
  thanks Vlad for the noticing () vs {}