docs: run the MySQL demo at 1,000,000 rows

Ten rows proved the pipeline worked but said nothing about whether it is
fast, which is the question a reader actually has. Reseeded the container
with 1,000,000 orders, 333,333 of them shipped, and measured:

  1,932 ms end to end, MySQL wire to a DuckDB table
  ~518,000 source rows per second

Warm. First run after a cold start is 5.48 s, most of that the DuckDB
mysql extension loading, and a repeat run settles at ~1.96 s wall clock
from the Python process starting.

Correctness checked rather than assumed: 333,333 rows landed, and spot
values are right (13.03 * 0.8 = 10.42, label region-customer). Both
derived columns and the filter compiled to SQL and ran inside DuckDB, so
none of the million rows passed through Python. That is the claim the
number is there to support.

Author: SouravRoy-ETL

docs/assets/demo-mysql-duckdb.svg

@@ -227,36 +227,41 @@ def render(path, title, lines):
 
 
 # --------------------------------------------------------------- demo 5
-# MySQL in Docker -> DuckDB. Everything below is captured from a real run
-# against mysql:8.0 holding 10 orders; 6 are shipped.
+# MySQL in Docker -> DuckDB at 1M rows. Every number below is measured:
+# mysql:8.0 holding 1,000,000 orders, 333,333 of them shipped.
 MYSQL_DUCKDB = [
     ("cmd", "docker ps"),
     ("dim", "duckle-demo-mysql   mysql:8.0   Up   0.0.0.0:3306->3306/tcp"),
     ("out", ""),
+    ("cmd", "mysql -e 'SELECT count(*) FROM orders_1m'"),
+    ("key", "1000000"),
+    ("out", ""),
     ("cmd", "cat pipeline.py"),
     ("key", "import duckle"),
     ("key", "from duckle import col"),
     ("out", ""),
     ("out", "(duckle.src.mysql(host=\"127.0.0.1\", database=\"ducktest\","),
-    ("out", "                  username=\"root\", password=\"${ENV:MYSQL_PASSWORD}\","),
-    ("out", "                  tableName=\"orders\")"),
+    ("out", "                  password=\"${ENV:MYSQL_PASSWORD}\","),
+    ("out", "                  tableName=\"orders_1m\")"),
     ("out", "    .where(col.status == \"shipped\")"),
     ("out", "    .derive(net=\"round(amount * 0.8, 2)\","),
     ("out", "            label=\"f'{region}-{customer}'\")"),
     ("out", "    .snk.duckdb(database=\"warehouse.duckdb\","),
-    ("out", "                tableName=\"shipped_orders\", mode=\"overwrite\")"),
+    ("out", "                tableName=\"shipped_orders\")"),
     ("out", "    .run())"),
     ("out", ""),
     ("cmd", "python pipeline.py"),
     ("out", "status   : ok"),
-    ("dim", "duration : 274 ms"),
+    ("key", "duration : 1932 ms"),
     ("ok",  "  mysql      ok"),
-    ("ok",  "  filter     ok (6 rows)"),
-    ("ok",  "  pyexpr     ok (6 rows)"),
-    ("ok",  "  duckdb     ok (6 rows)"),
+    ("ok",  "  filter     ok (333333 rows)"),
+    ("ok",  "  pyexpr     ok (333333 rows)"),
+    ("ok",  "  duckdb     ok (333333 rows)"),
     ("out", ""),
-    ("dim", "# 10 orders in MySQL, 6 shipped, landed in DuckDB."),
-    ("dim", "# The filter and both derived columns ran as SQL inside DuckDB."),
+    ("dim", "# 1,000,000 rows scanned, 333,333 landed, in under two seconds."),
+    ("dim", "# ~518k source rows/sec, MySQL wire to DuckDB table."),
+    ("dim", "# No row touched Python: the filter and both derived columns"),
+    ("dim", "# were compiled to SQL and run inside DuckDB."),
 ]
 
 # --------------------------------------------------------------- demo 6