Torch-snippets

☠️ Stop using rich's print and revert back to builtin print
🐞 Decouple AD and torch
🎉 Add a new chart - spider / radar

🧹 change code to remove future warnings in text_utils

🐞 AD string summary was buggy

🐞 all write modes are 'a' by default to avoid accidental overwriting
🧹 AD writes better string summary (support for multiline)

🧹 `print_module_ios_for` has better targeted functionality where you need to give submodules name
🎉 `clean_gpu_mem` and `get_latest_checkpoint` functions in torch_loader
🧹 `minor bugfix in AD`

🐞 minor bugfix in `AD2` where data keyword misbehaves
🧹 `video` has better size functionality
🎉 `if all is given in print_ios_for_module, all modules are printed`
🎉 `AD2.dict` is an alias for `AD2.to_dict`
🎉 better `AD2.summary` for pandas dataframes and `AD2.summary` respects max_items for keys as well
🎉 new alias `pd.read_pqt` for `pd.read_parquet`
🐞 wrap `tree` into python