[Misc] Load FP8 kv-cache scaling factors from checkpoints (#4893)

The 2nd PR for #4532.

This PR supports loading FP8 kv-cache scaling factors from a FP8 checkpoint (with .kv_scale parameter).

vllm/model_executor/models/opt.py

@@ -91,7 +91,8 @@ class OPTAttention(nn.Module):
         self.attn = Attention(self.num_heads,
                               self.head_dim,
                               scale=self.scaling,
-                              cache_config=cache_config)
+                              cache_config=cache_config,
+                              quant_config=quant_config)
 
     def forward(
         self,