Add OpProf campaign: protocols, results, patches, run evidence (P0-P6)

Workload-conditioned operator profiling on patched vLLM 0.24.0 +
Qwen3-30B-A3B/H20. H1b PASS (irregular patterns carry +23-45pp R64
raggedness, 8-45% token-efficiency loss vs rectangular controls);
mechanism decomposition kills the padding narrative and finds the
arrival-uniformization artifact (-12.9%); cross-version churn surface
shows TP2/MNS64 -29.4% across vLLM 0.20->0.24 while the argmax held.
Raw Layer-1 JSONL streams (507 MB) stay on disk, git-ignored; footer
sidecars and metrics are tracked.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

runs/opprof-phase6/phase6/cells/tp4_mns8/warmup/result.json

@@ -0,0 +1,74 @@
+{
+  "anchor": 0.016591107009,
+  "cell": "tp4_mns8",
+  "early_stop_reason": "",
+  "early_stopped": false,
+  "exact_output_count": 16,
+  "feasible": false,
+  "interval": {
+    "elapsed_s": 6.775592838,
+    "end_mono_ns": 201547098885716,
+    "end_wall_ns": 1783869974225312732,
+    "start_mono_ns": 201540323292878,
+    "start_wall_ns": 1783869967449719953
+  },
+  "invariants": {
+    "arrival_nondecreasing": true,
+    "exact_output_or_failed": true,
+    "outcomes_cover_selected": true,
+    "raw_lengths_present": true,
+    "selected_nonempty": true,
+    "warmup_16": true,
+    "warmup_exact_16": true,
+    "warmup_long": true
+  },
+  "kind": "warmup",
+  "mns": 8,
+  "observed_count": 16,
+  "pass_rate": 0.625,
+  "schema": 1,
+  "selection": {
+    "arrival_order_sha256": "99f2dcf01a54ab81776395fdef964b2a109d37fd8d44f971f3723e99a012b35e",
+    "arrival_s": {
+      "distinct_n": 16,
+      "finite_n": 16,
+      "max": 3.0116999999999736,
+      "min": 0.0,
+      "missing_n": 0,
+      "n": 16
+    },
+    "count": 16,
+    "long_gt4096": 5,
+    "offered_req_s": 0.26666666666666666,
+    "offered_req_s_per_gpu": 0.06666666666666667,
+    "raw_input_tokens": {
+      "distinct_n": 16,
+      "finite_n": 16,
+      "max": 8150.0,
+      "min": 810.0,
+      "missing_n": 0,
+      "n": 16
+    },
+    "raw_length_order_sha256": "3b3e3258b82a4168ea8c3836458779878ca106893e1af47b23576e89de4b245f",
+    "request_id_order_sha256": "e321d81e19450f59c2ac022f6ae487ec4c72024453d015376c2df60b97167d76"
+  },
+  "slo_pass_count": 10,
+  "study_sha256": "76e9d642ba9570917cfeeea848ed92f27904e87760eedabd48805b84e82eca26",
+  "tp": 4,
+  "tpot_ms": {
+    "distinct_n": 16,
+    "finite_n": 16,
+    "max": 18.231644558923787,
+    "min": 5.76333962215888,
+    "missing_n": 0,
+    "n": 16
+  },
+  "ttft_ms": {
+    "distinct_n": 16,
+    "finite_n": 16,
+    "max": 3136.144638003316,
+    "min": 1031.9823420140892,
+    "missing_n": 0,
+    "n": 16
+  }
+}