docs: update TODO — remove completed DeltaEncoding item, detail benchmark tasks

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

Author: dfa1

TODO.md

@@ -11,13 +11,18 @@
       and benchmark commit SHA so numbers don't rot silently.
 
 - [ ] **#13 Cascading compressor — remaining follow-ups**
-    - [ ] **DeltaEncoding wire-format fix**: refactor to emit `(bases child, deltas child)` matching
-      Rust's 2-child wire format (currently 1-buffer flat — incompatible with Rust reader, see commit 09685a2).
-      Unblocks cross-compat for delta cascades.
-    - [ ] **Cross-compat test**: `JavaWritesRustReadsIntegrationTest` with `allowedCascading=3`
-      confirms Rust JNI reader decodes the same OHLC data (gates DeltaEncoding wire-format fix).
-    - [ ] **Benchmark**: extend `RustVsJavaWriteBenchmark` with cascading-off vs cascading-on variants;
-      capture compressed-bytes ratio (Java cascade / JNI), write throughput, decode throughput.
+    - [ ] **Cross-compat test**: Add `JavaWritesRustReadsIntegrationTest` in the `integration` module.
+      Write OHLC data with `WriteOptions.cascading(3)` via Java writer; read it back with the Rust JNI
+      reader and assert values round-trip correctly. Same OHLC schema / dataset as `RustVsJavaWriteBenchmark`.
+    - [ ] **Benchmark — cascading write variants**: In `RustVsJavaWriteBenchmark`, add:
+        - `javaWriteCascading()` — same body as `javaWrite()` but use `WriteOptions.cascading(3)` instead of `WriteOptions.defaults()`
+        - After both `javaWrite` and `javaWriteCascading` run, log the compressed-bytes ratio
+          (`javaWriteCascading file size / javaWrite file size`) via a JMH `AuxCounters` or a
+          `@TearDown` print so it appears in the benchmark output.
+    - [ ] **Benchmark — cascading read variants**: In `RustVsJavaReadBenchmark`, add a
+      `javaReadCascading()` method that reads a file written with `WriteOptions.cascading(3)`.
+      Warm up the file in `@Setup`; measure decode throughput (rows/s). Compare against `javaRead()`
+      (no cascading) and `jniRead()` to show the cascading cost/benefit at read time.
 
 ## Tooling