Investigate whether dropping requests in another thread is faster

[prestwich]

• benchmark dropping Request and BatchRequest objects of various sizes with the following
  • drop(req),
  • std::thread::spawn(|| drop(req))
  • tokio::spawn(|| async move { drop req }
• Repeat benchmarks for
  • 1 KiB
  • 10 KiB
  • 100 KiB
  • 1 MiB
• Develop heuristic for whether dropping in another task is faster
• integrate heuristic into drops in Handler impls