[https://nvbugs/5478151][fix] Add missing spec for Llama-3.3 70B (#7267)

Signed-off-by: Balaram Buddharaju <[email protected]>
Co-authored-by: Larry <[email protected]>

Author: brb-nv

tests/integration/defs/accuracy/references/gpqa_diamond.yaml

@@ -5,6 +5,9 @@ meta-llama/Llama-3.3-70B-Instruct:
     accuracy: 45.55
   - quant_algo: FP8
     accuracy: 48.03
+  - quant_algo: FP8
+    kv_cache_quant_algo: FP8
+    accuracy: 48.03
 deepseek-ai/DeepSeek-R1:
   - quant_algo: NVFP4
     accuracy: 70.45