chore: ignore generated agent-bench output

Run archives, the aggregated report and the packaged standalone kit are all
regenerated from scripts; bench/runs alone is 16 MB across 1140 files.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

.gitignore

@@ -4,3 +4,10 @@ node_modules/
 *.log
 # Local research downloads and generation checkpoints
 /tmp/
+
+# Agent bench: regenerated from scripts, never edited by hand
+bench/runs/
+bench/report.json
+bench/hex-bench/
+bench/hex-bench.zip
+bench/pack/