chore(vllm): remove redundant nvlink-* dual composes (NVLink auto-detected) (#257)

The four dual/autoround-int4/nvlink-*.yml composes (nvlink-fp8-mtp,
nvlink-turbo, nvlink-dflash, nvlink-dflash-noviz) and their
vllm/dual-nvlink* launch slugs are removed. They were thin Docker Compose
`extends:` stubs whose ONLY override was NVLINK_MODE=force_on — fully
redundant since every base dual compose now auto-detects NVLink at boot
via detect_nvlink.sh (NVLINK_MODE=auto): it probes nvidia-smi and flips on
NCCL_P2P_LEVEL=NVL + custom-all-reduce when a bridge is present, else
NCCL_P2P_DISABLE=1 + --disable-custom-all-reduce. NVLink rigs get the
identical fast path from the base dual compose with no separate slug;
force it explicitly with `NVLINK_MODE=force_on scripts/switch.sh vllm/dual`
if auto-detect ever misses.

Scope (prune only; no engine-image change — the v0.22.0 stable-engine
consolidation is a separate follow-up):
- delete 4 nvlink-*.yml composes
- compose_registry.py: drop the 4 _entry blocks (50 -> 46 entries)
- test-compose-registry-disk.sh: count guards 50/51 -> 46/47
- profile_runtime.yml + patches.yml: drop the 4 nvlink blocks/slug refs
- test-profiles-compat.sh: retire the C13/E3 NVLink-required-compose
  scenarios (the estate NVLink-gating code is now dormant; cleanup tracked
  as a follow-up, code retained)
- docs: switch.sh help, AGENTS.md, HARDWARE.md, UPSTREAM.md, patches/README.md,
  fp8-mtp/dflash/dflash-noviz compose headers, per-model CHANGELOG removal
  entry; add a docs/README.md pointer to `switch.sh --list` as the
  authoritative registry-derived compose x slug matrix

Historical NVLink bench rows (JusefPol #31, danbedford #74/#92/#96) are
preserved in BENCHMARKS.md. Full test suite green (the lone
test-submit-bench failure is a worktree-isolation artifact — it needs
gitignored results/rebench/ fixtures absent from a fresh worktree; passes
on the working tree).

Co-authored-by: noonghunna <10742901+noonghunna@users.noreply.github.com>
Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

AGENTS.md

@@ -79,7 +79,7 @@ The directory hierarchy encodes model, engine, topology, and the weights artifac
 | `models/<model>/<engine>/` | Inference engine | `vllm` · `llama-cpp` · `sglang` |
 | `compose/<topology>/` | Hardware topology | `single` · `dual` · `multi3` · `multi4` · `multi8` |
 | `<quant>/` | Weights artifact / `weights_variant` slug | `autoround-int4` · `ubergarm-iq4ks` · `awq` |
-| `<serving>.yml` (filename) | Serving stack | `fp8-mtp.yml` · `turbo.yml` · `dflash.yml` · `nvlink-dflash-noviz.yml` |
+| `<serving>.yml` (filename) | Serving stack | `fp8-mtp.yml` · `turbo.yml` · `dflash.yml` · `int8.yml` |
 
 **Topology rule**: `single` (TP=1) and `dual` (TP=2) have no count ambiguity. `multi<N>` requires the count because N varies (3 / 4 / 5 / 6 / 8). Aligns with `docs/SINGLE_CARD.md` / `DUAL_CARD.md` / `MULTI_CARD.md` doc framing.
 
@@ -88,7 +88,7 @@ The directory hierarchy encodes model, engine, topology, and the weights artifac
 **Feature suffix order** (when stacking): interconnect → drafter → KV → vision modifier. Examples:
 - `dual/autoround-int4/turbo.yml` — TP=2 + AutoRound INT4 weights + TurboQuant KV
 - `dual/autoround-int4/dflash.yml` — TP=2 + AutoRound INT4 weights + DFlash drafter
-- `dual/autoround-int4/nvlink-dflash-noviz.yml` — TP=2 + NVLink + DFlash + no vision
+- _(the `nvlink-` interconnect prefix is reserved but currently unused — NVLink is auto-detected at boot via `NVLINK_MODE`, not encoded in the filename)_
 - `dual/autoround-int4/int8.yml` — TP=2 + AutoRound INT4 weights + INT8 PTH KV
 - `dual/awq/bf16-mtp.yml` — TP=2 + AWQ weights + BF16 KV + MTP
 - `multi4/autoround-int4/dflash.yml` — TP=4 + DFlash