feat: increase Llama max-model-len to 20480

Adjusted VRAM allocation for larger context window:
- Llama: 90% VRAM, 20480 context (up from 8192)
- BGE: 8% VRAM (down from 10%)

🤖 Generated with [Claude Code](https://claude.com/claude-code)

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

services/vllm/config_bge.yaml

@@ -2,6 +2,6 @@ model: BAAI/bge-large-en-v1.5
 host: "0.0.0.0"
 port: 8002
 uvicorn-log-level: "info"
-gpu-memory-utilization: 0.10
+gpu-memory-utilization: 0.08
 dtype: float16
 task: embed