What's wrong with this benchmark, the result is totally unexpected

``` group.bench_function("Flume (sync)", |b| { b.iter(|| { let (tx, rx) = flume::bounded(1024); let producer = thread::spawn(move || { for _ in 0..MSG_COUNT { tx.send(0).unwrap(); } }); let consumer = thread::spawn(move || { for _ in 0..MSG_COUNT { black_box(rx.recv().unwrap()); } }); producer.join().unwrap(); consumer.join().unwrap(); }); }); group.bench_function("Std MPSC (sync)", |b| { b.iter(|| { let (tx, rx) = std::sync::mpsc::sync_channel(1024); let producer = thread::spawn(move || { for _ in 0..MSG_COUNT { tx.send(0).unwrap(); } }); let consumer = thread::spawn(move || { for _ in 0..MSG_COUNT { black_box(rx.recv().unwrap()); } }); producer.join().unwrap(); consumer.join().unwrap(); }); }); ``` test result: ``` Benchmarking Queue Throughput/Flume (sync): Collecting 1000 samples in estimated 48.6 Queue Throughput/Flume (sync) time: [48.502 µs 48.558 µs 48.614 µs] thrpt: [21.064 Melem/s 21.088 Melem/s 21.113 Melem/s] change: time: [-52.993% -52.893% -52.790%] (p = 0.00 < 0.05) thrpt: [+111.82% +112.28% +112.73%] Performance has improved. Found 45 outliers among 1000 measurements (4.50%) 8 (0.80%) low severe 8 (0.80%) low mild 13 (1.30%) high mild 16 (1.60%) high severe Benchmarking Queue Throughput/Std MPSC (sync): Collecting 1000 samples in estimated 3 Queue Throughput/Std MPSC (sync) time: [33.888 µs 33.937 µs 33.994 µs] thrpt: [30.123 Melem/s 30.173 Melem/s 30.218 Melem/s] change: time: [-65.127% -65.036% -64.938%] (p = 0.00 < 0.05) thrpt: [+185.21% +186.00% +186.76%] Performance has improved. Found 77 outliers among 1000 measurements (7.70%) 7 (0.70%) low severe 22 (2.20%) low mild 35 (3.50%) high mild 13 (1.30%) high severe ```