feat(models): add Qwen3.5-122B-A10B-FP8 candidate

[krisztian-gajdar]

Adds Qwen/Qwen3.5-122B-A10B-FP8 to the existing SGLang adapter, pinned to Hugging Face revision a099dee70ccfcd8d5dda56aaa0b60cb8ecadabc9.

The candidate uses native FP8 weights, BF16 compute, tensor parallelism across two GPUs, and a bounded 32K non-thinking context with a 4096-token output cap. It includes an explicit rtx-pro-6000x2 profile and a matching non-speculative grammar profile. Tool parsing and sampling defaults follow the pinned model card.

Validation: canonical model configuration and adapter-option checks pass; catalog and tensor-parallel tests report 294 passed and 33 skipped. GPU acceptance remains pending, so this PR stays draft.

[coderabbitai]

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

• Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true

---

_{Comment @coderabbitai help to get the list of available commands.}