[Docs] Add Inkling to supported models - #55793

Purpose

InklingForCausalLM and InklingForConditionalGeneration are already registered in vllm/model_executor/models/registry.py (package vllm.models.inkling) with public HF weights, but docs/models/supported_models.md had no Inkling rows.

This adds:

• Text-only: InklingForCausalLM (LoRA ✅︎, PP ✅︎)
• Multimodal: InklingForConditionalGeneration with inputs T + I + A (image + audio; LoRA ✅︎, PP ✅︎)

Example HF models: thinkingmachines/Inkling, thinkingmachines/Inkling-NVFP4.

Local repro (no full clone)

HEAD=$(gh api repos/vllm-project/vllm/commits/main --jq .sha)

curl -fsSL "https://raw.githubusercontent.com/vllm-project/vllm/${HEAD}/vllm/model_executor/models/registry.py" | rg -n 'InklingFor(CausalLM|ConditionalGeneration)'
curl -fsSL "https://raw.githubusercontent.com/vllm-project/vllm/${HEAD}/vllm/models/inkling/nvidia/model.py" | rg -n 'class InklingFor|SupportsLoRA|SupportsPP|SupportsMultiModal|Only image or audio'
curl -fsSL "https://raw.githubusercontent.com/vllm-project/vllm/${HEAD}/tests/models/registry.py" | rg -n -A3 'InklingForCausalLM'
curl -fsSL "https://raw.githubusercontent.com/vllm-project/vllm/${HEAD}/docs/models/supported_models.md" | rg -n 'Inkling' || echo 'MISSING on main (expected before this PR)'

Test plan

• Confirm both new rows appear in the text-only and multimodal tables
• Confirm alphabetical placement (after HyperCLOVAX / Idefics3)
• Confirm LoRA/PP markers match SupportsLoRA / SupportsPP
• Confirm multimodal Inputs are T + I + A only