frank-primary: radiance native-FP8 TP=2 cutover (engine, payload build, rollback)

Replace the FreeToken PR587 endpoint (single-GPU: its TP=2 dies in the
pynccl.cu ROCm JIT) with the radiance engine on both R9700s, TP=2 via
libr4d's fused one-shot all-reduce. Official Qwen/Qwen3.8-Flash-Next-FP8
payload at native block-FP8 precision:

- scripts/build-frank-primary-rad.sh: staging (quantization_config drop =
  plugin SELECTION law) -> checkpoint reshape -> rad-convert with ONE
  --quant rule (PLE rtn fixed-scale); plan-only first.
- scripts/stack-qwen38-official-fp8.py: the two value-preserving reshapes
  (experts byte-stacked into the plugin's stacked layout; PLE's 128
  scaleless fp8 shards decoded to one bf16 shard with the checkpoint's own
  scale; the rtn rule re-encodes at that same scale - codes round-trip
  exactly and land the one-element scale plane the plugin declares).
- scripts/hfget.sh + verified download (131/131 shards sha256-match).
- froggeric v22.5 chat template vendored (--override-chat-template).
- serve script (pinned ledger: 2x40 GiB pools + 47.7 GiB PLE + prefix host),
  TTM preflight fixed for kernels with pages_limit only (no pages_max -
  an absent knob is skipped, not read as zero); pages_limit raised to
  52,428,800 pages = 200 GiB on the host pre-cutover.
- root unit, relinked; deployer gates image digest + .rad floor 158,000 MiB
  + health; prometheus re-adds the /metrics job.
- rollback: tag frank-primary-freetoken-pr587-20261005 (FreeToken trio +
  Dockerfile); payload + image stay on disk until the window closes.

deploy-remote.sh also carries the parallel session's frank-small payload
gates (swift15) - committed here per Frank's direction to commit the tree.

shared/prometheus/prometheus.yml

@@ -98,12 +98,13 @@ scrape_configs:
   # service down. Same host, same rule, so both local endpoints use the
   # gateway.
   #
-  # (2026-10-05) frank-primary left this file: FreeToken (the PR587 ROCm
-  # engine now serving it) has NO /metrics endpoint — the job scraped 404s
-  # forever, which is a structural never-green, not an outage. The
-  # netavark published-port quirk above is same-host only; plain LAN
-  # routing REACHES 192.168.0.235:7770 (health answers) — only the metrics
-  # path is absent. If FreeToken grows a /metrics, re-add the job here.
+  # (2026-10-05) frank-primary LEFT the llamacpp family with FreeToken
+  # (the PR587 ROCm engine then serving it had NO /metrics — the job
+  # scraped 404s forever, a structural never-green) and RE-JOINED the same
+  # day on the radiance engine: its exporter speaks vLLM-compatible names
+  # plus radiance:* families, a different namespace — hence the separate
+  # job below, not a re-add to the llamacpp pair. The netavark quirk above
+  # is same-host only; plain LAN routing REACHES 192.168.0.235:7770.
   - job_name: "frank-small"
     metrics_path: "/metrics"
     static_configs:
@@ -118,3 +119,15 @@ scrape_configs:
         labels:
           service: frank-large
 
+  # frank-primary (dankserver) — the radiance engine since 2026-10-05.
+  # A DIFFERENT host, so plain LAN routing is correct here (the
+  # host.containers.internal trick is same-host only): health answers at
+  # 192.168.0.235:7770 from this bridge container. Radiance exports
+  # vLLM-compatible names plus radiance:* families — NOT the llamacpp:*
+  # namespace, so llm-fleet rules must not be pointed at this target.
+  - job_name: "frank-primary"
+    metrics_path: "/metrics"
+    static_configs:
+      - targets: ["192.168.0.235:7770"]
+        labels:
+          service: frank-primary