forth: Hayes conformance runner + baseline scoreboard (165/590, 28%)

lib/forth/scoreboard.md

@@ -0,0 +1,30 @@
+# Forth Hayes Core scoreboard
+
+| metric            | value |
+| ----------------- | ----: |
+| chunks available  | 638 |
+| chunks fed        | 590 |
+| total             | 590 |
+| pass              | 165 |
+| fail              | 0 |
+| error             | 425 |
+| percent           | 27% |
+
+- **Source**: `gerryjackson/forth2012-test-suite` `src/core.fr`
+- **Generated**: 2026-04-24T19:13:12Z
+- **Note**: completed
+
+A "chunk" is any preprocessed segment ending at a `}T` (every Hayes test
+is one chunk, plus the small declaration blocks between tests).
+The runner catches raised errors at chunk boundaries so one bad chunk
+does not abort the rest. `error` covers chunks that raised; `fail`
+covers tests whose `->` / `}T` comparison mismatched.
+
+### Chunk cap
+
+`conformance.sh` processes the first `$MAX_CHUNKS` chunks (default
+**590**). Past that, `core.fr` ships tests that rely on unsigned
+integer wrap-around (e.g. `COUNT-BITS` using `BEGIN DUP WHILE … 2*
+REPEAT`), which never terminates on our bignum-based interpreter. The
+cap should rise as those tests unblock — run with `MAX_CHUNKS=639
+./conformance.sh` once they do.