Pymysql-pool

1. more complete custom query method db_query() and db_modify() within cursor object
2. add singleton pattern

1. Add more friendly methods of Cursor object: db_query() and db_modify().
2. Add logic to prevent reuse of a connection object that has already returned to the pool.
3. change pool.size to pool.available_num & pool.connection_num to pool.total_num.
4. More reasonable logger settings.

New designed pool model(use deque instead of queue.Queue), almost 20x performance when doing borrowing and returning actions.

Breaking changes: change the parameter `max_size` to `maxsize` in the init function of the ConnectPool class.

1. rewrite the init function of ConnectionPool:
1.1 remove hard limit of the pool
1.2 add arg `con_lifetime` for resolve server side close due to the 'wait_timeout' variable; and used for pool scalability
1.3 add arg `pre_create_num` indicate that is create some connection ad init phase
2. enhance the robustness of the 'put connection back to pool' function
3. expose a stats of used and available connections at the point, via the `connection_num` property