Prevent command out of sync errors with Prepared Statements

[sodabrew]

Thanks to @kamipo for diagnosing the problem and drafting the first PR #957

This fixes a regression caused by #912 due to calling rb_funcall between mysql_stmt_execute and mysql_stmt_store_result, it will cause mysql_stmt_close to be called in wrong order.

In further testing, rb_hash_aref can also call rb_funcall if the query_options hash has a default_proc set, so adding a stronger comment to remind future maintainers.

Fixes #956.

[sodabrew]

From the land of you must be f kidding me, prepared statement cursors disable microsecond timestamps. https://bugs.mysql.com/bug.php?id=86818

This is fixed in MariaDB and broken in MySQL 5.5, 5.6. Meanwhile, MySQL 5.7 and 8.0 just hang when cursors are enabled.

  1) Mysql2::Statement should prepare DateTime values with microseconds
     Failure/Error: expect(result.first['a'].strftime('%F %T.%5N %z')).to eql(now.strftime('%F %T.%5N %z'))

       expected: "2018-04-04 06:54:50.60401 -0700"
            got: "2018-04-04 06:54:51.00000 -0700"

       (compared using eql?)
     # ./spec/mysql2/statement_spec.rb:172:in `block (2 levels) in <top (required)>'

  2) Mysql2::Statement should prepare Time values with microseconds
     Failure/Error: expect(result.first['a'].strftime('%F %T.%5N %z')).to eql(now.strftime('%F %T.%5N %z'))

       expected: "2018-04-04 06:54:50.63292 -0700"
            got: "2018-04-04 06:54:51.00000 -0700"

       (compared using eql?)
     # ./spec/mysql2/statement_spec.rb:164:in `block (2 levels) in <top (required)>'