Run bench-taskset.sh through bash in bench-compare

The script is not executable in git, so executing it directly fails with
EACCES. Also show the benchmarked branch in the run name on push.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01UeniBAQefdj1j3odtudVTU
Signed-off-by: Claude <noreply@anthropic.com>

Author: claude

.github/workflows/bench-compare.yml

@@ -8,7 +8,7 @@
 
 name: Benchmark Compare
 run-name: >-
-  Bench compare ${{ inputs.head_ref || github.ref_name }} vs ${{ inputs.base_ref || 'develop' }}:
+  Bench compare ${{ inputs.head_ref || (github.event_name == 'push' && 'ji/duckdb-v2-scan-driver') || github.ref_name }} vs ${{ inputs.base_ref || 'develop' }}:
   ${{ inputs.benchmarks || 'test defaults' }}
 
 concurrency:

scripts/bench-compare.py

@@ -150,7 +150,7 @@ def run_one(exe: Path, filter: str | None, extra: list[str], pin: bool) -> dict[
     if filter:
         cmd.append(filter)
     if pin:
-        cmd = [str(SCRIPT_DIR / "bench-taskset.sh"), *cmd]
+        cmd = ["bash", str(SCRIPT_DIR / "bench-taskset.sh"), *cmd]
     print("+", " ".join(cmd), file=sys.stderr, flush=True)
     proc = subprocess.run(cmd, stdout=subprocess.PIPE, text=True, env={**os.environ, "RUST_BACKTRACE": "1"})
     sys.stderr.write(proc.stdout)