Added new commands to the CLI. The summary of the commands are as follows:

• clean: Removes files that don't match what nbautoexport would create based on current notebooks and current config. For example, the use case is removing stuff like Untitled.py to clean up after stuff from renamed or deleted notebooks.
• convert: This runs conversion with the configuration provided by CLI arguments (no .nbautoexport file needed).
• export: This runs conversion on a folder with a .nbautoexport configuration file or with override CLI options.
• install: The old main command of creating a .nbautoexport configuration file.

Additional changes:

• Adding an clean configuration option to automatically run the cleaning as part of post_save. Defaults to False.
• Add rst as an export format
• Add __main__.py to support for calling with python -m. This requires setting typer>=0.3.0.

Additional refactoring:

• Split source code into multiple modules
• Refactored export functionality into new function export.export_notebook to better support code reuse for convert and export commands.
• Refactored all instances of os.path to use pathlib
• Refactored .nbautoexport config to use pydantic model, with strict validation on fields

How stuff works:

• How clean, convert, and export know what files are notebooks:
  • New function utils.find_notebook will attempt to load every file in given directory with nbformat.
• How clean knows what files to expect:
  • New function clean.notebook_exports_generator holds logic for predicting exported files.
  • New function clean.get_extension grabs nbconvert exporters to figure out appropriate file extension

Fixed some bugs:

• The asciidoc, latex, markdown, and rst formats create separate directories for image files that weren't getting moved into the subfolders.
• pdf format resulted in a unicode decode error because the postprocessor was trying to read the pdf file back in as unicode text in order to remove the cell numbers
• nbconvert grabs CLI arguments from sys.argv, which leads to really weird errors if nbconvert gets initiated through something like pytest or nbautoexport. Previously this was worked around in the tests by monkeypatching sys.argv. Instead, we now use a context manager to temporarily clear sys.argv whenever nbconvert is initialized, which prevents errors no matter if it was nbautoexport, pytest, or jupyter that starts the outer process.

Looking for feedback:

• Are the names of the commands clear?
  • convert and export could be confusing.
  • install is what the original function for setting up was named this whole time. Are there better names? configure? watch? I personally like watch. Opened issue Split install command into install and configure #34
• Right now we only support organize_by of notebooks and extensions. Wondering if it would make sense to support None, i.e., keep converted output in parent directory and not in subfolders. Opened Proposal: Add support for having no subfolders #32

TODO:

• Update unit tests to work with refactor
• Add new unit tests for new commands

Known issues, thinking will save these for the future:

• Determining appropriate file extension to get expected export files is currently clunky for script. nbconvert actually determines the language from loading the .ipynb file and then dispatches to the specific language exporter. We may want to do the same thing. ended up implementing
• I don't actually know how to get the exporters for languages besides Python (expecting R and Julia?). The nbconvert library only directly supports python. We'll have to figure out how it knows about others. Currently this will return .txt. Ran this against an R notebook I had, and it looks up the .r from the notebook metadata under file_extension.
• The expected export files functionality does not support the subdirectories that asciidoc, markdown, and rst create for images, so clean will remove them. This may require figuring out naming conventions, and/or refactoring the expected_exports method to return globs instead of Paths. ended up implementing
• Functionality that searches for notebooks (clean, export, convert) currently just glob for *.ipynb files. This may miss stuff, and we should potentially instead try checking any file with nbformat instead. ended up implementing

Note: I tried to get the root of the CLI to point to install but it doesn't work, because other subcommands will conflict with arguments.

Resolves #20, #21