document why not having shims is good

[grosser]

I came here from rbenv/rbenv#1223

[peterb]

❤️

[postmodern]

Decided to mention in 29af09c and 4f38caa about when shims require constant regeneration, which is an annoying chore. I feel like rbenv could just optimize the shim regeneration or background the process and declare the problem solved.

[eregon]

Those commits don't mention the performance cost of shims though, so it seems a separate concern.
👍 for this PR.

[peterb]

Had another look at this and I've changed my mind.

@postmodern I think the time it takes for the shim to execute is dependent on the system architecture and might involve binaries? If I'm right, then it feels to me like this is something that would be better off treated as an add-on, maybe by providing a hook for plug-ins.

@eregon It would be nice to include some real-world figures in the docs, but surely that would have to mention the system specifications? The laws of nature and mathematics would tell us that a single number isn't statistically meaningful.