feat(bookslm): v2.2.0 - BooksLM chat AI contextuel par répertoire + 4 providers AI

BooksLM (#76):
- Panneau chat slide-in 450px, clic-droit répertoire → 🧠 BooksLM
- Collecte récursive .md avec limites (200 fichiers, 200K chars)
- Cache SHA-256, redaction secrets, priorité README/index
- SSE streaming, badges sources cliquables, historique localStorage
- Commandes palette: BooksLM ouvrir/nouvelle conversation

Providers AI (4 nouveaux):
- NVIDIA (integrate.api.nvidia.com)
- QwenCloud (dashscope.aliyuncs.com)
- Xiaomi (api.xiaomi.com)
- Mistral (api.mistral.ai)
- Tous OpenAI-compatible, auto-listing modèles

Fix: dropdown config-select suit maintenant le thème (option bg/color)
27 tests BooksLM + 466 tests au total

backend/bookslm_routes.py

@@ -0,0 +1,147 @@
+"""BooksLM API routes — directory-scoped AI chat for Obsidian vaults."""
+
+import json
+import logging
+from pathlib import Path
+
+from fastapi import APIRouter, Depends, HTTPException
+from fastapi.responses import StreamingResponse
+from pydantic import BaseModel, Field
+
+from backend.auth.middleware import check_vault_access, require_auth
+from backend.bookslm import build_system_prompt, collect_directory_context
+from backend.indexer import get_vault_data
+
+logger = logging.getLogger("obsigate.bookslm_routes")
+router = APIRouter(prefix="/api/ai/bookslm", tags=["BooksLM"])
+
+
+# ── Request models ──
+
+
+class BooksLMContextRequest(BaseModel):
+    vault: str = Field(description="Vault name")
+    directory: str = Field(default="", description="Relative directory path within the vault")
+
+
+class BooksLMChatRequest(BaseModel):
+    vault: str = Field(description="Vault name")
+    directory: str = Field(default="", description="Relative directory path within the vault")
+    message: str = Field(description="User message")
+    conversation_history: list[dict[str, str]] = Field(
+        default_factory=list,
+        description="Previous conversation turns [{role, content}]",
+    )
+
+
+# ── Endpoints ──
+
+
+@router.post("/context")
+async def api_bookslm_context(
+    req: BooksLMContextRequest,
+    current_user=Depends(require_auth),
+):
+    """Collect directory context for BooksLM.
+
+    Returns file list, content, and metadata for the specified directory.
+    """
+    if not check_vault_access(req.vault, current_user):
+        raise HTTPException(status_code=403, detail=f"Accès refusé à la vault '{req.vault}'")
+
+    vault_data = get_vault_data(req.vault)
+    if not vault_data:
+        raise HTTPException(status_code=404, detail=f"Vault '{req.vault}' not found")
+
+    vault_path = Path(vault_data["path"])
+    context = collect_directory_context(vault_path, req.directory)
+    return context
+
+
+@router.post("/chat")
+async def api_bookslm_chat(
+    req: BooksLMChatRequest,
+    current_user=Depends(require_auth),
+):
+    """Chat with AI about directory contents (BooksLM).
+
+    Builds context from the directory, then sends the user message
+    with a system prompt containing all file contents to the AI provider.
+    Returns an SSE stream with the response.
+    """
+    if not check_vault_access(req.vault, current_user):
+        raise HTTPException(status_code=403, detail=f"Accès refusé à la vault '{req.vault}'")
+
+    vault_data = get_vault_data(req.vault)
+    if not vault_data:
+        raise HTTPException(status_code=404, detail=f"Vault '{req.vault}' not found")
+
+    vault_path = Path(vault_data["path"])
+
+    # Collect context
+    context = collect_directory_context(vault_path, req.directory)
+    if context["file_count"] == 0:
+        raise HTTPException(status_code=404, detail="Aucun fichier markdown trouvé dans ce dossier")
+
+    # Build system prompt
+    system_prompt = build_system_prompt(context)
+
+    # Call AI provider
+    from backend.ai import DEFAULT_PROVIDER, PROVIDERS, _call_deepseek_openrouter, _call_gemini
+
+    # Build messages with conversation history
+    messages_text = ""
+    if req.conversation_history:
+        for turn in req.conversation_history:
+            role = turn.get("role", "user")
+            content = turn.get("content", "")
+            if role == "user":
+                messages_text += f"\n\nUtilisateur : {content}"
+            elif role == "assistant":
+                messages_text += f"\n\nAssistant : {content}"
+
+    # Current message
+    user_prompt = req.message
+    if messages_text:
+        user_prompt = f"Historique de la conversation :{messages_text}\n\nQuestion actuelle : {req.message}"
+
+    async def generate_sse():
+        try:
+            cfg_name = DEFAULT_PROVIDER
+            # Check if default provider is available
+            if cfg_name != "gemini" and cfg_name in PROVIDERS:
+                if not PROVIDERS[cfg_name].get("api_key"):
+                    # Find first available provider
+                    for pname, pcfg in PROVIDERS.items():
+                        if pcfg.get("api_key") and pname != "gemini":
+                            cfg_name = pname
+                            break
+
+            if cfg_name == "gemini" and PROVIDERS.get("gemini", {}).get("api_key"):
+                response = await _call_gemini(user_prompt, system_prompt, temperature=0.3, max_tokens=4096)
+            else:
+                response = await _call_deepseek_openrouter(
+                    user_prompt, system_prompt,
+                    provider=cfg_name if cfg_name in PROVIDERS else None,
+                    temperature=0.3,
+                    max_tokens=4096,
+                )
+
+            # Send the full response as a single SSE event
+            data = json.dumps({"token": response}, ensure_ascii=False)
+            yield f"event: message\ndata: {data}\n\n"
+            yield "event: done\ndata: {}\n\n"
+        except Exception as e:
+            logger.error(f"BooksLM chat error: {e}")
+            error_data = json.dumps({"error": str(e)}, ensure_ascii=False)
+            yield f"event: error\ndata: {error_data}\n\n"
+
+    return StreamingResponse(
+        generate_sse(),
+        media_type="text/event-stream",
+        headers={
+            "Cache-Control": "no-cache",
+            "Connection": "keep-alive",
+            "X-Accel-Buffering": "no",
+        },
+    )