Error details & Generator improvements

[sherbang]

So I started with some more improvements to providing error details in exceptions, then I got carried away and did a bunch of work on the generator.

Changes include:

• Add more checks to ruff (and related code changes)
• Parsing the error that Manage returns when you try to create an object that already exists
• Moving generator templates out into separate files so code highlighting works
• Move generator out of pyconnectwise namespace, so it isn't included with the shipped library
• Document how to run the generator
• Glued together the different parts of the generator, so it all runs in one go (model gen, schema cleanup, endpoint gen)
• Changed line-length in our tooling to match what the generator puts out
• Make sure that we don't generate endpoints that refer to models that don't exist (fail faster)

I think that's everything worth mentioning. Now I've got to figure out/remember what started me on this journey in the first place. 😁

[sherbang]

Nice automation!

[sherbang]

I pushed a few more commits today. The most notable one is lazy-loading endpoints.

Before the lazy-loading endpoints change:
python -c 'import pyconnectwise' would take 2.8 seconds to run and use ~240Mb of memory.

After the change:
python -c 'import pyconnectwise' takes 136 milliseconds to run and use ~50Mb of memory.

[Yoshify]

This PR is fantastic - I love the additions to lazy loading the imports. I still have more testing I'd like to do with it (with generating Automate files especially) but other than that I can't see any downsides to this. I expect to merge soon.

[Yoshify]

LGTM

[Yoshify]

Merged - thanks again for your incredible contribution!