feat: alora adapters for ollama

[aanokh]

Pull Request

Issue

No dedicated issue, requested by Gabe Goodhart in the Mellea Slack thread on Ollama aLoRA support, building on Simra Faisal's earlier prototype after the Epic #929 refactor.

Description

Adds adapter function support to OllamaModelBackend, so helpers like
core.check_certainty work against an Ollama model that bundles an aLoRA
adapter (for example gabegoodhart/granite4.1-uncertainty:3b). Ollama activates the
adapter when its trigger text appears in the prompt, and the io.yaml rewriter
already appends that trigger, so the backend just mixes in AdapterMixin and
routes Intrinsic and ALoraRequirement actions to a new
_generate_from_intrinsic (inspired by the OpenAI backend), which sends the rewritten
conversation through Ollama chat with logprobs
enabled.

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)

Verified live on Ollama 0.33.1.

Attribution

• AI coding assistants used (IBM Bob and Claude Code)

---

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

If your PR adds or modifies one of the types below, check the matching box. A checklist of type-specific review items will be posted as a comment.

• Component
• Requirement
• Sampling Strategy
• Tool

[jakelorocco]

I've kicked off the acceptance tests; we may need to alter our test setup scripts to download different ollama models to get them to run though.

[planetf1]

Suggest we continue with this after we can rebase on #1619 which completes the adapter refactor - happy to complete the task if @aanokh isn't available.

[planetf1]

Superseded by #1634, which rebases and takes over this work on the current adapter API. The original authored commit is retained in the replacement branch history.