feat(memory): integrate memory system with context engine (#97)

## Changes

### New Context Type
- Add MEMORY to ContextType enum for agent memory context
- Create MemoryContext class with subtypes (working, episodic, semantic, procedural)
- Factory methods: from_working_memory, from_episodic_memory, from_semantic_memory, from_procedural_memory

### Memory Context Source
- MemoryContextSource service fetches relevant memories for context assembly
- Configurable fetch limits per memory type
- Parallel fetching from all memory types

### Agent Lifecycle Hooks
- AgentLifecycleManager handles spawn, pause, resume, terminate events
- spawn: Initialize working memory with optional initial state
- pause: Create checkpoint of working memory
- resume: Restore from checkpoint
- terminate: Consolidate working memory to episodic memory
- LifecycleHooks for custom extension points

### Context Engine Integration
- Add memory_query parameter to assemble_context()
- Add session_id and agent_type_id for memory scoping
- Memory budget allocation (15% by default)
- set_memory_source() for runtime configuration

### Tests
- 48 new tests for MemoryContext, MemoryContextSource, and lifecycle hooks
- All 108 memory-related tests passing
- mypy and ruff checks passing

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

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

backend/app/services/context/budget/allocator.py

@@ -30,6 +30,7 @@ class TokenBudget:
     knowledge: int = 0
     conversation: int = 0
     tools: int = 0
+    memory: int = 0  # Agent memory (working, episodic, semantic, procedural)
     response_reserve: int = 0
     buffer: int = 0
 
@@ -60,6 +61,7 @@ class TokenBudget:
             "knowledge": self.knowledge,
             "conversation": self.conversation,
             "tool": self.tools,
+            "memory": self.memory,
         }
         return allocation_map.get(context_type, 0)
 
@@ -211,6 +213,7 @@ class TokenBudget:
                 "knowledge": self.knowledge,
                 "conversation": self.conversation,
                 "tools": self.tools,
+                "memory": self.memory,
                 "response_reserve": self.response_reserve,
                 "buffer": self.buffer,
             },
@@ -264,9 +267,10 @@ class BudgetAllocator:
             total=total_tokens,
             system=int(total_tokens * alloc.get("system", 0.05)),
             task=int(total_tokens * alloc.get("task", 0.10)),
-            knowledge=int(total_tokens * alloc.get("knowledge", 0.40)),
-            conversation=int(total_tokens * alloc.get("conversation", 0.20)),
+            knowledge=int(total_tokens * alloc.get("knowledge", 0.30)),
+            conversation=int(total_tokens * alloc.get("conversation", 0.15)),
             tools=int(total_tokens * alloc.get("tools", 0.05)),
+            memory=int(total_tokens * alloc.get("memory", 0.15)),
             response_reserve=int(total_tokens * alloc.get("response", 0.15)),
             buffer=int(total_tokens * alloc.get("buffer", 0.05)),
         )
@@ -317,6 +321,8 @@ class BudgetAllocator:
             budget.conversation = max(0, budget.conversation + actual_adjustment)
         elif context_type == "tool":
             budget.tools = max(0, budget.tools + actual_adjustment)
+        elif context_type == "memory":
+            budget.memory = max(0, budget.memory + actual_adjustment)
 
         return budget
 
@@ -338,7 +344,7 @@ class BudgetAllocator:
             Rebalanced budget
         """
         if prioritize is None:
-            prioritize = [ContextType.KNOWLEDGE, ContextType.TASK, ContextType.SYSTEM]
+            prioritize = [ContextType.KNOWLEDGE, ContextType.MEMORY, ContextType.TASK, ContextType.SYSTEM]
 
         # Calculate unused tokens per type
         unused: dict[str, int] = {}