Stop an empty env var from shadowing .env, and refuse a total OCR blackout

Two faults found by running the pipeline inside the Docker image for the first
time, both of which let a 25-page run finish "successfully" having stored nothing.

The image declares ENV DASHSCOPE_API_KEY= so the variable is visible to
docker inspect and overridable at docker run. That leaves it set-but-empty in
every container, and load_dotenv used os.environ.setdefault, so the empty string
won and the mounted .env was ignored. An empty value now counts as unset; a real
value still wins, because overriding on the command line is the point.

The consequence was worse than a missing key. Every page failed, the per-page
fallback kept the font-mangled embedded text -- exactly the text OCR is there to
replace -- and the run went on to spend a full detection pass on it. Tolerating
one unreadable page is right; tolerating every page is a configuration error
wearing a success message, so Document now raises OcrUnavailable when OCR was
attempted and not one page came back, quoting the first real failure. A page
range that needed no OCR at all is still not an error.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Ci5T3gWDhhmqDec3e1XtYZ

leanbench/config.py

@@ -17,7 +17,15 @@ DEFAULT_API_KEY_ENV = "DASHSCOPE_API_KEY"
 
 
 def load_dotenv(path: Path | str = ".env") -> None:
-    """Load ``KEY=VALUE`` lines into os.environ without overwriting what is set."""
+    """Load ``KEY=VALUE`` lines into os.environ without overwriting a real value.
+
+    A variable that is set but *empty* counts as unset. This is not pedantry: the
+    Docker image declares ``ENV DASHSCOPE_API_KEY=`` so the variable is documented
+    and overridable at ``docker run``, which left it present-and-empty in every
+    container. With plain ``setdefault`` that empty string shadowed the .env file,
+    and the only symptom was every OCR page failing with "No API key" -- a whole
+    25-page run spent on text the pipeline had already judged unreadable.
+    """
     path = Path(path)
     if not path.is_file():
         return
@@ -26,7 +34,10 @@ def load_dotenv(path: Path | str = ".env") -> None:
         if not line or line.startswith("#") or "=" not in line:
             continue
         key, _, value = line.partition("=")
-        os.environ.setdefault(key.strip(), value.strip().strip('"').strip("'"))
+        key = key.strip()
+        if os.environ.get(key):  # a non-empty value wins; empty or absent does not
+            continue
+        os.environ[key] = value.strip().strip('"').strip("'")
 
 
 def _env(name: str, default: str) -> str: