chore(context): refactor for consistency, optimize formatting, and simplify logic

- Cleaned up unnecessary comments in `__all__` definitions for better readability.
- Adjusted indentation and formatting across modules for improved clarity (e.g., long lines, logical grouping).
- Simplified conditional expressions and inline comments for context scoring and ranking.
- Replaced some hard-coded values with type-safe annotations (e.g., `ClassVar`).
- Removed unused imports and ensured consistent usage across test files.
- Updated `test_score_not_cached_on_context` to clarify caching behavior.
- Improved truncation strategy logic and marker handling.

backend/app/services/context/compression/truncation.py

@@ -78,7 +78,7 @@ class TruncationStrategy:
         )
 
     @property
-    def TRUNCATION_MARKER(self) -> str:
+    def truncation_marker(self) -> str:
         """Get truncation marker from settings."""
         return self._settings.truncation_marker
 
@@ -141,7 +141,9 @@ class TruncationStrategy:
             truncated_tokens=truncated_tokens,
             content=truncated,
             truncated=True,
-            truncation_ratio=0.0 if original_tokens == 0 else 1 - (truncated_tokens / original_tokens),
+            truncation_ratio=0.0
+            if original_tokens == 0
+            else 1 - (truncated_tokens / original_tokens),
         )
 
     async def _truncate_end(
@@ -156,17 +158,17 @@ class TruncationStrategy:
         Simple but effective for most content types.
         """
         # Binary search for optimal truncation point
-        marker_tokens = await self._count_tokens(self.TRUNCATION_MARKER, model)
+        marker_tokens = await self._count_tokens(self.truncation_marker, model)
         available_tokens = max(0, max_tokens - marker_tokens)
 
         # Edge case: if no tokens available for content, return just the marker
         if available_tokens <= 0:
-            return self.TRUNCATION_MARKER
+            return self.truncation_marker
 
         # Estimate characters per token (guard against division by zero)
         content_tokens = await self._count_tokens(content, model)
         if content_tokens == 0:
-            return content + self.TRUNCATION_MARKER
+            return content + self.truncation_marker
         chars_per_token = len(content) / content_tokens
 
         # Start with estimated position
@@ -188,7 +190,7 @@ class TruncationStrategy:
             else:
                 high = mid - 1
 
-        return best + self.TRUNCATION_MARKER
+        return best + self.truncation_marker
 
     async def _truncate_middle(
         self,
@@ -201,7 +203,7 @@ class TruncationStrategy:
 
         Good for code or content where context at boundaries matters.
         """
-        marker_tokens = await self._count_tokens(self.TRUNCATION_MARKER, model)
+        marker_tokens = await self._count_tokens(self.truncation_marker, model)
         available_tokens = max_tokens - marker_tokens
 
         # Split between start and end
@@ -218,7 +220,7 @@ class TruncationStrategy:
             content, end_tokens, from_start=False, model=model
         )
 
-        return start_content + self.TRUNCATION_MARKER + end_content
+        return start_content + self.truncation_marker + end_content
 
     async def _truncate_sentence(
         self,
@@ -236,7 +238,7 @@ class TruncationStrategy:
 
         result: list[str] = []
         total_tokens = 0
-        marker_tokens = await self._count_tokens(self.TRUNCATION_MARKER, model)
+        marker_tokens = await self._count_tokens(self.truncation_marker, model)
         available = max_tokens - marker_tokens
 
         for sentence in sentences:
@@ -248,7 +250,7 @@ class TruncationStrategy:
                 break
 
         if len(result) < len(sentences):
-            return " ".join(result) + self.TRUNCATION_MARKER
+            return " ".join(result) + self.truncation_marker
         return " ".join(result)
 
     async def _get_content_for_tokens(