[docs] Attention backends + continuous batching

[stevhliu]

Refactors the optimization docs, starting with the attention backend and continuous batching docs (more updates to come in subsequent PRs!).

• reorg the toctree to create a clearer inference optimization section for inference (adds the AttentionInterface, continuous batching, and kernels guides)
• rename the "LLMs" section into a more generic "Generate API" section especially since we have VLMs now
• add ContinuousMixin and ContinuousBatchingManager to the API reference
• polish the Attention Interface doc (table describing the available backends, using kernels, etc.) - maybe you can review this section @Cyrilvallez 🙏
• polish the Continuous batching doc (break up the ContinuousBatchingManager example to better show the different options for using it, adds clearer PagedAttention and sliding window attention sections, high-level description of how the mixin, manager, and scheduler work together, move the monitoring section to the transformers serve docs) - maybe you can review this section @McPatate 🙏

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.