"op read" is pretty slow, ~700ms per invocation

Hi all, I've noticed that op read is pretty slow, taking on the order of 700ms per invocation. Here's a benchmark using the https://github.com/sharkdp/hyperfine tool: % hyperfine --warmup...