feat(constants): add llm-d-vllm framework label

[ezrasilvera]

What

Registers a new framework llm-d-vllm in FW_REGISTRY (packages/constants) with the display label "llm-d vLLM".

Why

We are adding llm-d (vLLM) as a new benchmark framework in the InferenceX benchmark framework - GB200 DeepSeek-V4-Pro FP4 P/D-disaggregated recipes (mid-curve 1P1D at conc 256/512/1024 and high-tpt 2P1D at conc 4096). Once those
sweeps ingest, the result rows carry framework: llm-d-vllm.

Without this entry the dashboard would not treat it as a canonical framework (FRAMEWORK_KEYS is derived from FW_REGISTRY) and would render the series with the uppercased slug "LLM-D-VLLM" instead of "llm-d vLLM".

Change

• packages/constants/src/framework-aliases.ts: add 'llm-d-vllm': { label: 'llm-d vLLM' } to FW_REGISTRY (placed
  alphabetically, after dynamo-vllm).
• packages/constants/src/framework-aliases.test.ts: add a matching
  FRAMEWORK_LABELS assertion for the new label.

---

Note

Low Risk
Read-only display and caching changes with a backward-compatible cache key bump; no auth or data-mutation paths.

Overview
Adds llm-d-vllm as a canonical framework ("llm-d vLLM") and wires it through the Current InferenceX Image flow so new P/D-disaggregated benchmark rows display and filter correctly.

The latest-images API/query now returns disagg from configs; the UI uses that flag for Single node vs Disaggregated filtering instead of inferring disagg from dynamo-* / mori-* prefixes. baseFramework is centralized in latest-image-utils (including llm-d-vllm → vLLM for framework filters and upstream release lookup). Copy/tooltips mention llm-d among disaggregated stacks; the API cache key is bumped to latest-images-v2 so clients pick up the new payload shape.

^{Reviewed by Cursor Bugbot for commit 3ea6cc5. Bugbot is set up for automated code reviews on this repo. Configure here.}

[vercel]

@ezrasilvera is attempting to deploy a commit to the SemiAnalysisAI Team on Vercel.

A member of the Team first needs to authorize it.

[adibarra]

ideally, should be merged right before SemiAnalysisAI/InferenceX#2050