Textllm

- Made specifying a filename optional. Instead, will create a new file. If an *existing* directory is specified, will create the new file there.
- `--rename` defaults to True if a new file is created. Still defaults to `$TEXTLLM_AUTO_RENAME` otherwise which itself defaults to False
- Removed `--create` setting. Specifying a new file is all that is needed.
- Removed the double `--edit` where it opened up again. I think it's not worth it. Better to just have the user submit again.

- Changed default to `--stream`
- Added a tester. It's not perfect but it covers most cases

Minor:

- More robust parsing of settings

Made it so if you specify `--edit --edit`, it will create a recursive loop calling the editor after each chat too. This makes it *almost* like an interactive chat. To exit, just exit the editor without saving. Note that `--prompt` is only applied on the first round.

Also some documentation tweaks.

- Added `--stream / --no-stream` to stream the model output to stdout in addition to the file.
- Added `--prompt` and `--edit` flags edit or add a new prompt.

The two above flags make it possible to treat textllm as a simple client without even opening the file!

Minor:

- Fixed bug that made textllm still rename a file.
- Decided to *remove* all of the commented models. They are added to the readme.

More commented models in the template. Move to different numbering scheme

20250201.0.xx

These are the initial release but also in flux (the different .xx) as I work out kinks. Still. **BETA**