Add BACKGROUND_MODEL to split chat and background LLM calls

Dossier extraction and compaction now use a separate BACKGROUND_MODEL
(dolphin-mistral) to avoid Qwen3.5 timeout issues on background tasks.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

bot.py

@@ -26,6 +26,7 @@ DISCORD_TOKEN = os.environ["DISCORD_TOKEN"]
 LITELLM_BASE_URL = os.environ["LITELLM_BASE_URL"]
 LITELLM_API_KEY = os.getenv("LITELLM_API_KEY", "anything")
 MODEL = os.getenv("MODEL", "gpt-4")
+BACKGROUND_MODEL = os.getenv("BACKGROUND_MODEL", MODEL)
 SYSTEM_PROMPT = os.getenv("SYSTEM_PROMPT", "You are a helpful assistant.")
 MAX_CONTEXT = int(os.getenv("MAX_CONTEXT", "20"))
 MCP_URL = os.getenv("MCP_URL", "http://mcp-searxng:4001/mcp")
@@ -147,7 +148,7 @@ async def maybe_compact_dossier(key: str, name: str):
         try:
             response = await asyncio.wait_for(
                 llm.chat.completions.create(
-                    model=MODEL,
+                    model=BACKGROUND_MODEL,
                     messages=[
                         {
                             "role": "system",
@@ -236,7 +237,7 @@ async def extract_and_update_dossier(user: discord.Member | discord.User, conver
         t0 = time.monotonic()
         try:
             response = await asyncio.wait_for(
-                llm.chat.completions.create(model=MODEL, messages=extraction_messages),
+                llm.chat.completions.create(model=BACKGROUND_MODEL, messages=extraction_messages),
                 timeout=BACKGROUND_LLM_TIMEOUT,
             )
             elapsed = time.monotonic() - t0
@@ -301,7 +302,7 @@ async def extract_and_update_self_dossier(conversation: list[dict]):
         t0 = time.monotonic()
         try:
             response = await asyncio.wait_for(
-                llm.chat.completions.create(model=MODEL, messages=extraction_messages),
+                llm.chat.completions.create(model=BACKGROUND_MODEL, messages=extraction_messages),
                 timeout=BACKGROUND_LLM_TIMEOUT,
             )
             elapsed = time.monotonic() - t0