Improvements for maintainability

[amrit110]

vector-inference is starting to expand to more models and has onboarded users. I'm looking to contribute by helping improve its maintainability. Concretely, I have 4 tasks in mind:

• Reduce maintenance overhead by having dependabot submit security updates and dependency updates as long as they conform to the requirements of the library
• Add a docs build to create a webpage to help users
• Improve the dev build tooling (migrate to uv from poetry)
• Add unit tests for the python code in the vec-inf package.
• Automate docker image build process and pushing to dockerhub/GCR repo

[amrit110]

#23 for dependabot configuration

[amrit110]

#28 for migrating to uv

[amrit110]

#30 for docs build

[jwilles]

One issue that I will add for discussion - VI is slow to onboard new models because of the strong coupling between code and configuration. We currently bundle the job config with the VI package. This means that users need update their VI package to add a few lines to their models.csv which doesn't feel great.