Let names and assertion messages carry what comments were carrying

Renames so the code states its own invariants: _SINGLETON_ROW_ID for the
one-row version table, get_tag_origins_by_reference so the inner mapping's
direction is in the name, _ROWS_PER_TRANSACTION for what the batch size bounds.

In the tests, a named helper replaces the explanation of why the engine is
shared across sessions, a named constant replaces the empty-string "any unique
hash" sentinel, and each test's claim moves into its assertion message, where
whoever's build just went red will actually read it.

app/assets/database/queries/__init__.py

@@ -60,7 +60,7 @@ from app.assets.database.queries.semantics import (
     bulk_set_loader_paths,
     get_file_backed_references_page,
     get_semantics_version,
-    get_tags_by_reference,
+    get_tag_origins_by_reference,
     set_semantics_version,
 )
 from app.assets.database.queries.tags import (
@@ -122,7 +122,7 @@ __all__ = [
     "get_references_by_paths_and_asset_ids",
     "get_references_for_prefixes",
     "get_semantics_version",
-    "get_tags_by_reference",
+    "get_tag_origins_by_reference",
     "get_unenriched_references",
     "get_unreferenced_unhashed_asset_ids",
     "insert_reference",

app/assets/database/queries/semantics.py

@@ -19,22 +19,22 @@ from app.assets.database.queries.common import (
 from app.assets.database.queries.tags import ensure_tags_exist
 from app.assets.helpers import get_utc_now
 
-_VERSION_ROW_ID = 1
+_SINGLETON_ROW_ID = 1
 
 AUTOMATIC_TAG_ORIGIN = "automatic"
 
 
 def get_semantics_version(session: Session) -> int:
-    row = session.get(AssetSemanticsVersion, _VERSION_ROW_ID)
+    row = session.get(AssetSemanticsVersion, _SINGLETON_ROW_ID)
     return int(row.version) if row is not None else 0
 
 
 def set_semantics_version(session: Session, version: int) -> None:
-    row = session.get(AssetSemanticsVersion, _VERSION_ROW_ID)
+    row = session.get(AssetSemanticsVersion, _SINGLETON_ROW_ID)
     if row is None:
         session.add(
             AssetSemanticsVersion(
-                id=_VERSION_ROW_ID, version=int(version), updated_at=get_utc_now()
+                id=_SINGLETON_ROW_ID, version=int(version), updated_at=get_utc_now()
             )
         )
     else:
@@ -93,7 +93,7 @@ def get_file_backed_references_page(
     ]
 
 
-def get_tags_by_reference(
+def get_tag_origins_by_reference(
     session: Session, reference_ids: list[str]
 ) -> dict[str, dict[str, str]]:
     if not reference_ids:

app/assets/semantics/reproject_derived.py

@@ -29,7 +29,7 @@ from app.assets.database.queries.semantics import (
     bulk_remove_automatic_tags,
     bulk_set_loader_paths,
     get_file_backed_references_page,
-    get_tags_by_reference,
+    get_tag_origins_by_reference,
 )
 from app.assets.helpers import normalize_tags
 from app.assets.semantics.step import InterruptCheck, SemanticsStepInterrupted
@@ -41,7 +41,7 @@ from app.assets.services.path_utils import (
 )
 from app.database.db import create_session
 
-_BATCH_SIZE = 500
+_ROWS_PER_TRANSACTION = 500
 
 
 @dataclass
@@ -86,7 +86,7 @@ def reproject_derived_state(
 
         with create_session() as session:
             rows = get_file_backed_references_page(
-                session, after_id=after_id, limit=_BATCH_SIZE
+                session, after_id=after_id, limit=_ROWS_PER_TRANSACTION
             )
             if not rows:
                 return summary
@@ -101,7 +101,9 @@ def _reproject_batch(
     vocabulary: set[str],
     summary: ReprojectionSummary,
 ) -> None:
-    stored_tags = get_tags_by_reference(session, [row.reference_id for row in rows])
+    stored_tags = get_tag_origins_by_reference(
+        session, [row.reference_id for row in rows]
+    )
 
     loader_paths: dict[str, str | None] = {}
     tags_to_add: list[tuple[str, str]] = []