Implemented a MediaGroupBuilder class in 'aiogram/utils/media_group.py' to help construct media groups. The class supports addition of different media types (audio, photo, video, document) to the media group with a maximum limit of 10 files. The functionality is demonstrated and usage is documented in 'docs/utils/media_group.rst'. Added related test cases in 'tests/test_utils/test_media_group.py'. This is to streamline and simplify the process of media group creation
* Ensure base type validation can handle non-dictionary values
The update introduces a condition to verify whether the values being validated are a dictionary before attempting to handle UNSET_TYPE in the aiogram base type. This adjustment helps to prevent potential errors or incorrect validation when non-dictionary values are faced.
* Added a test case for non-dictionary input in remove_unset method
* Added changelog
* Fixed tests
* PoC: Mount objects to the Bot instance, bind shortcuts to configured instance
* Fixe docstring of the bind method
* Pass Bot instance explicitly to the URLInputFile
* Added tests
* Added changelog
* Refactor aiogram client and update tests
Refactored base.py to improve code readability by separating response_type operation from model_validate(). Also, adjusted the parameters in URLInputFile() within test_input_file.py for better test coverage. Updated input_file.py to streamline read method and avoid unnecessary instantiation of Bot class. Lastly, adjusted typing in methods/base.py to enhance code clarity.
* Update changelog
Upgrade architecture + 5.0 Bot API (#469)
* Moved `methods`, `types` and `client` to root package
* Removed update handler from routers to dispatcher
* Reworked events propagation mechanism to handlers
* Reworked inner middlewares logic (very small change)
* Updated to Bot API 5.0
* Initial migration from MkDocs to Sphinx + config for readthedocs
Resolves#302. We decided to use sentinel pattern (https://python-patterns.guide/python/sentinel-object/) as a solution, but got a few problems with plain `object()`, so instead we use unittest.mock.sentinel and we hope it won't cause side effects.
Most of work done via tg-codegen (https://github.com/aiogram/tg-codegen/pull/1), so it's good to review only implementation of sentinel, processing sentinel in `prepare_parse_mode`, changes in base method model and little test fixes.
