chore: simplify redundant dict comprehensions (C416, C420)

Replace `{k: v for k in iter}` with `dict.fromkeys(iter, v)` (C420, 4x)
and `{k: v for k, v in zip(a, b)}` with `dict(zip(a, b))` (C416, 4x).

These are pure idiom cleanups flagged by ruff's flake8-comprehensions
rules. `dict.fromkeys` and `dict()` are marginally faster and more
readable, with identical behavior. Verified: same keys, same values,
same order.

Files touched:
- m_flow/eval/report.py
- m_flow/ingestion/documents/classify_documents.py
- m_flow/retrieval/episodic/bundle_search.py
- m_flow/retrieval/utils/fine_grained_triplet_search.py
- m_flow/retrieval/utils/procedural_bundle_search.py
- m_flow/tests/test_relational_db_migration.py

Author: alexzhu0

m_flow/eval/report.py

@@ -120,7 +120,7 @@ def _analyze_failures(
     ) -> tuple[List[FailureCase], Dict[str, int]]:
         """Analyze failure samples and bucket them."""
         failures = []
-        buckets: Dict[str, int] = {k: 0 for k in FAILURE_BUCKETS}
+        buckets: Dict[str, int] = dict.fromkeys(FAILURE_BUCKETS, 0)
 
         for case, result in zip(cases, results):
             if not result.ok:

m_flow/ingestion/documents/classify_documents.py

@@ -33,33 +33,16 @@
 }
 
 # Office document formats
-_OFFICE_EXTS = {ext: UnstructuredDocument for ext in ["docx", "doc", "odt", "xls", "xlsx", "ppt", "pptx", "odp", "ods"]}
+_OFFICE_EXTS = dict.fromkeys(["docx", "doc", "odt", "xls", "xlsx", "ppt", "pptx", "odp", "ods"], UnstructuredDocument)
 
 # Image formats
-_IMAGE_EXTS = {
-    ext: ImageDocument
-    for ext in [
-        "png",
-        "dwg",
-        "xcf",
-        "jpg",
-        "jpx",
-        "apng",
-        "gif",
-        "webp",
-        "cr2",
-        "tif",
-        "bmp",
-        "jxr",
-        "psd",
-        "ico",
-        "heic",
-        "avif",
-    ]
-}
+_IMAGE_EXTS = dict.fromkeys(
+    ["png", "dwg", "xcf", "jpg", "jpx", "apng", "gif", "webp", "cr2", "tif", "bmp", "jxr", "psd", "ico", "heic", "avif"],
+    ImageDocument,
+)
 
 # Audio formats
-_AUDIO_EXTS = {ext: AudioDocument for ext in ["aac", "mid", "mp3", "m4a", "ogg", "flac", "wav", "amr", "aiff"]}
+_AUDIO_EXTS = dict.fromkeys(["aac", "mid", "mp3", "m4a", "ogg", "flac", "wav", "amr", "aiff"], AudioDocument)
 
 # Merged mapping table
 _DOC_TYPE_MAP = {**_TEXT_EXTS, **_OFFICE_EXTS, **_IMAGE_EXTS, **_AUDIO_EXTS}

m_flow/retrieval/episodic/bundle_search.py

@@ -338,7 +338,7 @@ async def search_in_collection(collection_name: str):
 
     results = await asyncio.gather(*[search_in_collection(c) for c in cfg.collections])
 
-    node_distances = {c: r for c, r in zip(cfg.collections, results)}
+    node_distances = dict(zip(cfg.collections, results))
     edge_distances = node_distances.get("RelationType_relationship_name", None)
 
     return node_distances, edge_distances

m_flow/retrieval/utils/fine_grained_triplet_search.py

@@ -252,7 +252,7 @@ async def search_in_collection(collection_name: str):
             f"Vector collection retrieval completed: Retrieved distances from {sum(1 for res in results if res)} collections in {vector_collection_search_time:.2f}s"
         )
 
-        node_distances = {collection: result for collection, result in zip(collections, results)}
+        node_distances = dict(zip(collections, results))
 
         edge_distances = node_distances.get("RelationType_relationship_name", None)
 

m_flow/retrieval/utils/procedural_bundle_search.py

@@ -286,7 +286,7 @@ async def search_col(col: str):
         if all(not r for r in results):
             return ([], []) if return_bundles else []
 
-        node_distances: Dict[str, list] = {c: r for c, r in zip(collections, results)}
+        node_distances: Dict[str, list] = dict(zip(collections, results))
         edge_distances = node_distances.get("RelationType_relationship_name")
 
         # Apply bonuses

m_flow/tests/test_relational_db_migration.py

@@ -20,7 +20,7 @@
 
 
 def _nodes_dict(nodes):
-    return {n_id: data for (n_id, data) in nodes}
+    return dict(nodes)
 
 
 def _normalize_name(name: str) -> str: