fix(hf): handle ModelOption.THINKING chat templates

[planetf1]

Pull Request

Issue

Fixes #1212

Description

LocalHFBackend previously removed ModelOption.THINKING before applying the
tokenizer chat template. This change detects the template's supported thinking
variable (think, thinking, or enable_thinking) and forwards boolean
values under that name. It also canonicalises native aliases during option
resolution, so per-call options correctly override backend defaults.

This PR deliberately does not add Granite 4.2 model identifiers or change
defaults; #1587 owns that work and its GPU span coverage. The tokenizer-only
integration test uses the public Granite 4.2 3B ID directly to prove the
generic mapping works with the template that first exposes a thinking control.

During validation, the parallel tool-call regression tests were found to create
a live Ollama backend despite exercising only formatter behaviour. This PR
switches them to the existing mocked backend fixture so the local suite does
not require Ollama for those tests.

Testing

• Tests added to the respective file if code was changed
• New code has 100% coverage if code was added
• Ensure existing tests and github automation passes (a maintainer will kick off the github automation when the rest of the PR is populated)
  • uv run pytest test/backends/test_huggingface_filter_options.py -q: 33 passed, 3 skipped; Granite 4.2 tokenizer regression executed
  • uv run pytest test/backends/test_acall_tools_parallel_execution.py -q: 4 passed without Ollama
  • uv run ruff format --check . && uv run ruff check .: passed
  • uv run mypy .: passed
  • uv run pytest test/ -m "not qualitative": 3,988 passed, 179 skipped, 128 deselected, 1 xpassed

Attribution

• AI coding assistants used

---

Adding a new component, requirement, sampling strategy, or tool?

• Component
• Requirement
• Sampling Strategy
• Tool

NOTE: Please ensure you have an issue that has been acknowledged by a core contributor and routed you to open a pull request against this repository. Otherwise, please open an issue before continuing with this pull request.

[jakelorocco]

agree with the approach; I think there's a few small concerns

[jakelorocco]

This only forwards the value if it's a bool. That seems to differ slightly from the docstring updates, and I'm not sure it's the correct behavior.

[jakelorocco]

Can you please include a comment here that this is using model_options and specifically not using the backend_opts that have the THINKING flags removed?

[jakelorocco]

This being after the backend specificity flag also causes ordering differences between the other backends and hf. For instance:

dict = {ModelOption.THINKING: True, "thinking": False}

# Ollama
backend_dict = {"thinking": True}  # Since _make_backend_specific_and_remove prefers the ModelOption sentinel

# HF
backend_dict = {"thinking": False} # Since _make_backend_specific_and_remove gets rid of the THINKING sentinel; causing the regular "thinking" option to take precedence when looking if it exists in the `backend_opts` dict