Add README.

README.md

@@ -0,0 +1,29 @@
+# Assignment
+
+To launch, install Docker or Podman and run:
+
+```bash
+$ cargo test
+    Finished `test` profile [unoptimized + debuginfo] target(s) in 0.70s
+     Running unittests src/lib.rs (target/debug/deps/supermetal_assignment-a8159c3e18c41f11)
+
+running 1 test
+(took 2197.842923 ms)
+test test::test_mysql ... ok
+
+test result: ok. 1 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out; finished in 29.89s
+...
+```
+
+## Future direction
+
+There are a few obvious places to go with this, and you probably don't really
+care about these, but just so you know *I* know...:
+
+* Nothing is configurable outside of code. (E.g. the Parquet batch size is a constant.) Might have been better to do a CLI rather than a test. Production code would of course be more configurable/tunable.
+* My dataset is a bit lopsided--an employees table with 500K records and a departments table with only 9. This made it a bit difficult to definitively establish how much concurrency actually benefitted in some areas but intuitively it made sense, and would probably bear out on actual data.
+* As previously stated, I'm not familiar with Parquet/Arrow outside of my use of
+  it with DuckDB/Overture Maps. There are probably additional ways to speed
+  things up, and I look forward to learning about those when we work together.
+
+Thanks, and do let me know if you have any additional questions.