feat(providers): register GPT-5.6 models (sol, terra, luna)

[chinmay1819]

What

Adds gpt-5.6-sol, gpt-5.6-terra and gpt-5.6-luna to the OpenAI provider's
bundled registry (src/skillspector/providers/openai/model_registry.yaml) and
to the root model_registry.yaml.

Why

These models already worked through SKILLSPECTOR_MODEL, but because they
weren't in the registry, SkillSpector:

• fell back to a 128K context window, so large files were split into far more
  LLM chunks than needed
• logged a "not found in model_registry.yaml" warning on every scan
• refused to start when SKILLSPECTOR_STRICT_MODEL_VALIDATION=true

How

Registry entries only, no code changes. All three models get the values from
OpenAI's model docs:

Model	context_length	max_output_tokens
gpt-5.6-sol	1,050,000	128,000
gpt-5.6-terra	1,050,000	128,000
gpt-5.6-luna	1,050,000	128,000

• Sol caps input at 922K tokens. SkillSpector budgets 75% of the context
  window (~787K), which stays under that cap, so no extra field is needed.
• Prices appear only as YAML comments with the source link. SkillSpector
  still never computes cost, as described in docs/INFERENCE_USAGE.md.
• DEFAULT_MODEL stays gpt-5.4.

Testing

• test_metadata_known_model now checks the context window and max output
  tokens for all three models.
• make test-unit, make lint and make format-check all pass.
• A --no-llm scan with strict model validation starts cleanly for each model.