this fixes subprogs missing keys

[jhheider]

closes #1085

[mxcl]

Nice, I had a similar patch coming.

[mxcl]

Again: unbelievable I didn’t spot this before! How on earth so many things I tested worked fine I have no idea.

[jhheider]

i hate the two for-loops, but didn't feel like messing around with maps and zips and so forth.

[jhheider]

my guess is because it's a script running under an execve that running in an interpreter... etc. levels on levels.

[mxcl]

Hilariously a couple hours ago I introduced a new test that exposes it: https://github.com/pkgxdev/pkgx/actions/runs/12835540088/job/35795376923?pr=1084#step:19:9

[mxcl]

i hate the two for-loops, but didn't feel like messing around with maps and zips and so forth.

Seems fine. We're iterating over two different things. Not sure there's a easy way to zip them together and do it clearly.

[jhheider]

joyous. i guess the "right" way to do it is to clone std::env::vars(), then loop through input and insert_or_join each key in a single loop.

[jhheider]

couldn't live with two loops. clone env, prepend needed keys.

[mxcl]

Definitely need some unit tests for these pieces

[jhheider]

that'd be an interesting one to spot regardless. needs a smart test. but preventing regressions should definitely be possible. an integration test stands a better chance.