Torch-snippets

🎉 `tryy` a new decorator to wrap try except with optional default return as well as print stack trace if need be
✨ `Timer` can given time-remaning approximation based on both instantaneous or average speed
✨ `logging_mode` functions can be used as both context managers and decorators
✨ better formatting for `now()`
✨ `show` auto generates a title if possible
✨ `show` function includes a `framecount` parameter to specify which frame of the stack it should search for the title. See its usage in `show_big_dataframe`
✨ `yolo_2_df` and `df_2_yolo` functions in adapters

✨ `AD` let's you know if a collection is list, set or tuple
✨ NEW DEFAULT: use `info` column to show text else `text`
🐞 fix a bug where `P` misbehaves in `AD` when it doesn't exist
🐞 fix a bug in `show` tables
🐞 `logger_mode` context will gracefully exit
✨ `to` is compatible with `transformers.BatchFeature`
⏮️ Revert back `P` as the code is incompatible with py312

🆕 `parse`ing a .ini file will add a `META` key by default which has `file` and `name` information. This header can be used in the file variables without creating META to begin with
🆕 `parse` will merge a base config if a `META.base` key is found. Works recursively on base files (base of base etc...)
🆕 `P` can tab complete existing file path name and access that file as an attribute
💡 Did you know parse can perform arithmetic?
✨ `registry` and `parse` are imported by default
✨ `jitter` can work on `pd.DataFrame` directly
🔨 `bbs2df` clips negative coordinates
✨ Using 🏷️ to denote data-type in `AD`
🐞 `AD2` is `AD` now
🐞 `AD` preserves dict order during serialization
✨ `warn_on_fail` decorator gives you a free try catch
✨ `show`ing a dataframe of the correct format inside and outside jupyter environments
🐞 `AD` has improved `get`, `update`, `map` and new `write_config`, `trymap` and `__json__`
✨ `P` has two new properties `.sz` for size of file and `.tree` to print the tree

🐞 `AD` hotfix

🎉 `store_scrap` is a new way to store on disk and show jupyter cell outputs in other notebooks.
Best for presenting complex analyses without worrying about running time-consuming notebooks
🎉 Add support for `P` in AD `summary` and `write_json`
🎉 `__json__` supports custom objects' serializability
🎉 `write_json` is compatible with above feature
🎉 `AD_MAX_ITEMS` if given as -1 will change it to 1000
🐞 `iou` will parse input dataframes more gracefully
🐞 `AD` minor bug fix
🎉 `tree` has a better default
🎉 New functions `folder_structure_to_dict` and `folder_structure_to_json` in `paths`
🎉 Add `jitter` (int) argument to `show` so that bounding boxes can be a bit jittered
🎉 Add support for changing `spinner` in `notify_waiting`
🎉 `dumpdill` can print a custom message (see `store_scrap` in paths.py)

🎉 AD `__contains__` can do a nested `in` 'x.y.z' in AD(x={'y': {'z': 10}}) == True