feat: compare safetensors and gguf on cpu and gpu

.scratch/distributed-gguf-runtime/evidence/DGR-001/README.md

@@ -11,9 +11,11 @@
 
 - Locks the DGR-001 benchmark contract in code.
 - Pins the architecture-aligned baseline to **DeepSeek-V2-Lite-Chat** (`deepseek2`).
-- Uses the smallest DeepSeek-family GGUF target selected for this story: **Q2_K** via `second-state/DeepSeek-V2-Lite-Chat-GGUF`.
+- Uses the same model on both sides of the comparison:
+  - **safetensors:** `deepseek-ai/DeepSeek-V2-Lite-Chat` in **BF16**
+  - **GGUF:** `second-state/DeepSeek-V2-Lite-Chat-GGUF` in **Q2_K**
 - Exposes a machine-readable JSON contract with:
-  - benchmark lanes for `transformers` safetensors and `llama.cpp` GGUF
+  - benchmark lanes for `transformers` safetensors and `llama.cpp` GGUF on **CPU** and **GPU**
   - concurrency levels `1` and `4`
   - the required metrics list
   - an explicit stop condition for “no meaningful speed or fit benefit”