Provide default value for y axis limit

A previous change incorrectly made ceiling a required variable, breaking
my Docker deployment, which relied on the default auto-ranging behavior.
This commit makes this argument optional, as intended.

Bonus changes:
* Pipe stdout and stderr of `crossword` to make it a little easier to
debug issues from logs.
* Add a reference to remind myself how to actually use Docker.

cloud_run.py

@@ -62,7 +62,12 @@ def generate_plot():
 
 
 def update_csv():
-    subprocess.run(["crossword", LOCAL_CSV_FILENAME], check=True)
+    subprocess.run(
+        ["crossword", LOCAL_CSV_FILENAME],
+        check=True,
+        stdout=subprocess.PIPE,
+        stderr=subprocess.PIPE,
+    )
 
 
 @app.route("/")

docker.md

@@ -0,0 +1,21 @@
+# Docker + Cloud Run
+
+Documenting my janky workflow.
+
+Image name: `us.gcr.io/<project id>/<image name>`
+
+```sh
+# Refresh credentials
+gcloud auth login
+
+# Building an image
+docker build -t `<image name>` .
+
+# Open a shell in container so you can run commands and test
+docker run -it --rm us.gcr.io/xword-stats/xword:latest bash
+
+# Push image to registry
+docker push us.gcr.io/xword-stats/xword:latest
+
+# Use Cloud console to point Cloud Run to latest image
+```

plot/plot.py

@@ -153,10 +153,10 @@ def save_split_vln_plot(df, out_path, ymax):
     plt.close()
 
 
-def generate(in_file, out_file, ceiling):
+def generate(in_file, out_file, ceiling = None):
     df = parse_data(in_file)
 
-    if ceiling == 0:
+    if ceiling is None:
         # Pick an appropriate y-axis, balancing being robust to outliers vs. showing all data
         ymax = df["solve_time_secs"].quantile(0.99) / 60
     else: