refactor(agents): move MAC middleware impls out of shared kernel

knowledge_search, memory_injection and scoped_model_fallback no longer
belong in the cross-agent kernel (app/agents/shared/middleware): they are
consumed only inside multi_agent_chat. Relocate each impl next to the
builder that uses it:

- knowledge_search.py -> multi_agent_chat/shared/middleware/ (genuinely
  shared: its _render_priority_message feeds kb_context_projection, used by
  both the main agent and the KB subagent)
- memory_injection.py -> multi_agent_chat/shared/middleware/ (beside its
  memory.py builder)
- scoped_model_fallback.py -> multi_agent_chat/shared/middleware/resilience/
  (beside fallback.py/bundle.py)

Impls moved verbatim (git rename). Builders/consumers now import the local
sibling; main_agent knowledge_priority imports the new shared path; shared
middleware barrel trimmed.

Tests: repoint imports; convert the knowledge_search monkeypatch targets
from brittle dotted-string form to object-based patching (monkeypatch.setattr
on the imported module), which is robust to import ordering. No behavior
change.

surfsense_backend/app/agents/multi_agent_chat/main_agent/middleware/knowledge_priority.py

@@ -4,8 +4,10 @@ from __future__ import annotations
 
 from langchain_core.language_models import BaseChatModel
 
+from app.agents.multi_agent_chat.shared.middleware.knowledge_search import (
+    KnowledgePriorityMiddleware,
+)
 from app.agents.shared.filesystem_selection import FilesystemMode
-from app.agents.shared.middleware import KnowledgePriorityMiddleware
 from app.services.llm_service import get_planner_llm
 
 

surfsense_backend/app/agents/multi_agent_chat/shared/middleware/kb_context_projection.py

@@ -12,9 +12,10 @@ from langgraph.runtime import Runtime
 from app.agents.multi_agent_chat.shared.state.filesystem_state import (
     SurfSenseFilesystemState,
 )
-from app.agents.shared.middleware.knowledge_search import _render_priority_message
 from app.utils.perf import get_perf_logger
 
+from .knowledge_search import _render_priority_message
+
 _perf_log = get_perf_logger()
 
 

surfsense_backend/app/agents/multi_agent_chat/shared/middleware/memory.py

@@ -2,9 +2,10 @@
 
 from __future__ import annotations
 
-from app.agents.shared.middleware import MemoryInjectionMiddleware
 from app.db import ChatVisibility
 
+from .memory_injection import MemoryInjectionMiddleware
+
 
 def build_memory_mw(
     *,

surfsense_backend/app/agents/multi_agent_chat/shared/middleware/resilience/bundle.py

@@ -12,13 +12,13 @@ from langchain.agents.middleware import (
 
 from app.agents.shared.feature_flags import AgentFeatureFlags
 from app.agents.shared.middleware import RetryAfterMiddleware
-from app.agents.shared.middleware.scoped_model_fallback import (
-    ScopedModelFallbackMiddleware,
-)
 
 from .fallback import build_fallback_mw
 from .model_call_limit import build_model_call_limit_mw
 from .retry import build_retry_mw
+from .scoped_model_fallback import (
+    ScopedModelFallbackMiddleware,
+)
 from .tool_call_limit import build_tool_call_limit_mw
 
 

surfsense_backend/app/agents/multi_agent_chat/shared/middleware/resilience/fallback.py

@@ -5,11 +5,11 @@ from __future__ import annotations
 import logging
 
 from app.agents.shared.feature_flags import AgentFeatureFlags
-from app.agents.shared.middleware.scoped_model_fallback import (
-    ScopedModelFallbackMiddleware,
-)
 
 from ..flags import enabled
+from .scoped_model_fallback import (
+    ScopedModelFallbackMiddleware,
+)
 
 
 def build_fallback_mw(

surfsense_backend/app/agents/shared/middleware/__init__.py

@@ -9,20 +9,12 @@ from app.agents.shared.middleware.kb_persistence import (
     KnowledgeBasePersistenceMiddleware,
     commit_staged_filesystem_state,
 )
-from app.agents.shared.middleware.knowledge_search import (
-    KnowledgePriorityMiddleware,
-)
-from app.agents.shared.middleware.memory_injection import (
-    MemoryInjectionMiddleware,
-)
 from app.agents.shared.middleware.permission import PermissionMiddleware
 from app.agents.shared.middleware.retry_after import RetryAfterMiddleware
 
 __all__ = [
     "BusyMutexMiddleware",
     "KnowledgeBasePersistenceMiddleware",
-    "KnowledgePriorityMiddleware",
-    "MemoryInjectionMiddleware",
     "PermissionMiddleware",
     "RetryAfterMiddleware",
     "SurfSenseCompactionMiddleware",