Prelapsarian require latency capture

[countvajhula]

Summary of Changes

This PR will not be merged. It's just here to measure the require time latency on an older "prelapsarian" state of the code (i.e. right before "the fall" of the require latency 🐶 ), so that we can add an accurate data point in the tracked benchmark data. The data can be found in each run of the benchmarks job for this branch, in the JSON output in the "run benchmark" step.

This branch is based on 3483112, although I'm not sure if the GitHub UI makes that clear anywhere.

FYI @benknoble 🙂

[countvajhula]

The results from five runs are:

{"name":"(require qi)","unit":"ms","value":490.0}
{"name":"(require qi)","unit":"ms","value":410.00000000000006}
{"name":"(require qi)","unit":"ms","value":490.0}
{"name":"(require qi)","unit":"ms","value":459.99999999999994}
{"name":"(require qi)","unit":"ms","value":570.0}

The average is (~> (490 410 490 460 570) (-< + count) / round) = 484ms.
For good measure, by throwing out the two extreme values, it's (~> (490 490 460) (-< + count) / round) = 480ms

[countvajhula]

And for completeness the baseline "postlapsarian" value, five runs on the merged commit that removed the requires: 2ebb865

{"name":"(require qi)","unit":"ms","value":220.0}
{"name":"(require qi)","unit":"ms","value":170.00000000000003}
{"name":"(require qi)","unit":"ms","value":220.00000000000003}
{"name":"(require qi)","unit":"ms","value":160.00000000000003}
{"name":"(require qi)","unit":"ms","value":169.99999999999997}

Same calculations:

> (~> (220 170 220 160 170) (-< + count) / round)
188
> (~> (220 170 170) (-< + count) / round)
187

[countvajhula]

Data added: https://countvajhula.github.io/qi/benchmarks/