feat(context): enhance performance, caching, and settings management

- Replace hard-coded limits with configurable settings (e.g., cache memory size, truncation strategy, relevance settings).
- Optimize parallel execution in token counting, scoring, and reranking for source diversity.
- Improve caching logic:
  - Add per-context locks for safe parallel scoring.
  - Reuse precomputed fingerprints for cache efficiency.
- Make truncation, scoring, and ranker behaviors fully configurable via settings.
- Add support for middle truncation, context hash-based hashing, and dynamic token limiting.
- Refactor methods for scalability and better error handling.

Tests: Updated all affected components with additional test cases.

backend/app/services/context/config.py

@@ -104,9 +104,21 @@ class ContextSettings(BaseSettings):
         le=1.0,
         description="Compress when budget usage exceeds this percentage",
     )
-    truncation_suffix: str = Field(
-        default="... [truncated]",
-        description="Suffix to add when truncating content",
+    truncation_marker: str = Field(
+        default="\n\n[...content truncated...]\n\n",
+        description="Marker text to insert where content was truncated",
+    )
+    truncation_preserve_ratio: float = Field(
+        default=0.7,
+        ge=0.1,
+        le=0.9,
+        description="Ratio of content to preserve from start in middle truncation (0.7 = 70% start, 30% end)",
+    )
+    truncation_min_content_length: int = Field(
+        default=100,
+        ge=10,
+        le=1000,
+        description="Minimum content length in characters before truncation applies",
     )
     summary_model_group: str = Field(
         default="fast",
@@ -128,6 +140,12 @@ class ContextSettings(BaseSettings):
         default="ctx",
         description="Redis key prefix for context cache",
     )
+    cache_memory_max_items: int = Field(
+        default=1000,
+        ge=100,
+        le=100000,
+        description="Maximum items in memory fallback cache when Redis unavailable",
+    )
 
     # Performance settings
     max_assembly_time_ms: int = Field(
@@ -165,6 +183,28 @@ class ContextSettings(BaseSettings):
         description="Minimum relevance score for knowledge",
     )
 
+    # Relevance scoring settings
+    relevance_keyword_fallback_weight: float = Field(
+        default=0.5,
+        ge=0.0,
+        le=1.0,
+        description="Maximum score for keyword-based fallback scoring (when semantic unavailable)",
+    )
+    relevance_semantic_max_chars: int = Field(
+        default=2000,
+        ge=100,
+        le=10000,
+        description="Maximum content length in chars for semantic similarity computation",
+    )
+
+    # Diversity/ranking settings
+    diversity_max_per_source: int = Field(
+        default=3,
+        ge=1,
+        le=20,
+        description="Maximum contexts from the same source in diversity reranking",
+    )
+
     # Conversation history settings
     conversation_max_turns: int = Field(
         default=20,
@@ -253,11 +293,15 @@ class ContextSettings(BaseSettings):
             "compression": {
                 "threshold": self.compression_threshold,
                 "summary_model_group": self.summary_model_group,
+                "truncation_marker": self.truncation_marker,
+                "truncation_preserve_ratio": self.truncation_preserve_ratio,
+                "truncation_min_content_length": self.truncation_min_content_length,
             },
             "cache": {
                 "enabled": self.cache_enabled,
                 "ttl_seconds": self.cache_ttl_seconds,
                 "prefix": self.cache_prefix,
+                "memory_max_items": self.cache_memory_max_items,
             },
             "performance": {
                 "max_assembly_time_ms": self.max_assembly_time_ms,
@@ -269,6 +313,13 @@ class ContextSettings(BaseSettings):
                 "max_results": self.knowledge_max_results,
                 "min_score": self.knowledge_min_score,
             },
+            "relevance": {
+                "keyword_fallback_weight": self.relevance_keyword_fallback_weight,
+                "semantic_max_chars": self.relevance_semantic_max_chars,
+            },
+            "diversity": {
+                "max_per_source": self.diversity_max_per_source,
+            },
             "conversation": {
                 "max_turns": self.conversation_max_turns,
                 "recent_priority": self.conversation_recent_priority,