Roadmap for v0.2

[abetlen]

• Migrate project to scikit-build-core #489
• Simplify dev / local setup using pyproject.toml and makefile #490
• Use numpy arrays for LogitsProcessors and StopCriteria to avoid copies / allocations #491
• Llama2 #488
• Configurable chat templates #492
• Add support for OpenAI-style functions #494
• Expose scores and input_ids in Llama model #493
• Add batched inference #771
• Speculative sampling #675