Add HuggingFace Transformers model support

[Copilot]

Adds a GIMKit adapter for HuggingFace transformers models, mirroring the pattern of existing adapters (vLLM, OpenAI).

Changes

• src/gimkit/models/transformers.py: New Transformers class extending outlines.models.transformers.Transformers with GIMKit's __call__ interface, plus from_transformers(model, tokenizer) factory function
• pyproject.toml: Added transformers>=4.50.0 as an optional dependency ([transformers] extra)
• src/gimkit/models/__init__.py, src/gimkit/__init__.py: Exported from_transformers

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer
from gimkit import from_transformers, guide

model_name = "Qwen/Qwen2.5-1.5B-Instruct"
hf_model = AutoModelForCausalLM.from_pretrained(model_name)
hf_tokenizer = AutoTokenizer.from_pretrained(model_name)

model = from_transformers(hf_model, hf_tokenizer)
result = model(f"The capital of France is {guide(name='city')}.")
print(result.tags["city"].content)  # e.g. "Paris"

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.