Move some parts of the engine into modules that can be installed separately

Nobody has claimed this yet.

I've been working on an application for the Objection Engine that doesn't require things like sentiment analysis or specialized Unicode parsing, but I've been struggling a bit with getting it installed on my server because of those packages.

This is a screenshot from a fresh virtual environment, with only Objection Engine (and the dependencies it has) installed:

Torch and the NVIDIA drivers(?) together take up over 3 GB, and the `.venv` folder as a whole takes up about 4 GB; by contrast, the actual assets folder that Objection Engine uses is 133 MB.

Based on this, I'm wondering if it may make sense to look into splitting Objection Engine into separate modules that can be installed separately or all together, depending on the user's needs. For example, in my use case because I don't need the machine learning capabilities, I could install *only* the rendering engine.

To maintain compatibility, this repository could serve as the "all together" option, with the core rendering engine being split off into its own repository/module that is required and imported by this one.

No files, tests, or entry points are named. Start by mapping the package's current dependencies and imports, then define separate installable modules for the rendering engine and optional sentiment-analysis, Unicode-parsing, and machine-learning components; done means rendering-only installation works while the existing all-in-one repository remains compatible.

Written by the indexing model from the issue text.