While not stickily necessary for most classes (since user isn't directly creating them) it still unifies our approach.
However for some like ReplyKeyboardHide where users are making the classes themselves it should improve IDE autocomplete support.
* Properly split and handle arguments in CommandHandler
* Update the docstring for pass_args in CommandHandler
* Properly split and handle arguments in StringCommandHandler
* bot.py: add create_references method
* create bot reference in webhook handler, use create_references on new updates
* message.py: implement reply_text
* echobot2.py: use Message.reply_text
* fix create_references in webhook handler
* add some more instance methods
* Chat.kick_member and unban_member
* bot.py: Create bot references in outgoing messages
* add tests for everything testable
* test_updater.py: add create_references method to MockBot
* remove Bot.create_references and refactor TelegramObject.de_json to take the additional parameter bot
* List bot as named kwarg where used
* file.py: Use Bot.request property instead of Bot._request attr
Should close#400.
* Add parse_entity
* Add parse_entities
* Add MessageEntity types as constants to MessageEntity.
* Add MAX_MESSAGE_ENTITIES to constants.py
Note: the value has been found by experimentation as opposed to extracted from the api docs.
* Add tests for parse_entity and parse_entities
* Create a Request class which maintains its own connection pool
* When creating a Bot instance a new Request instance will be created if one wasn't supplied.
* Updater is responsible for creating a Request instance if a Bot instance wasn't provided.
* Dispatcher: add method to run async functions without decorator
* Dispatcher can now run as a singleton (allowing run_async decorator to work) as it always did and as multiple instances (where run_async decorator will raise RuntimeError)
