modules/musl-cross: use echo from the PATH to support NixOS

[rbreslow]

NixOS doesn't have a traditional FHS where echo is available at /bin/echo. Instead, we must rely on the PATH for any distribution-managed utilities. Reverses #106.

---

Testing

See: #1282.

[tlaurion]

@rbreslow Can you make sure that .circleci/config.yaml is present in your PRs?
That way, CircleCI will build your commit, making sure that it passes compilation without regression for compilation prior of merging.

(Weird since on my side, I have to delete .circleci/config.yaml when I do not want my commits to be compiled by CircleCI)

Please force push this PR (and all others)!
Otherwise I will have to create another branch, cherry-pick your changes (making me co-author) and merge those instead of these. This will facilitate testing and collaboration. Note that you can (should) create an account on CircleCI and make it follow your Heads repo. That way, any commit you do on github's Heads fork would be compiled on your CircleCI instance automatically, and only PRs would be compiled per osresearch's instance (Open Source Tier, not a problem for now but might become one, we never know when service agreement will change).

Thanks!

[rbreslow]

Yes, no worries. I noticed this as well and had to sign into CircleCI and authorize my fork of Heads. Now it looks like things are working.

[tlaurion]

@rbreslow This build is failing on CircleCI for hopefully a transcient situation that won't happen again (your two other PRs didn't fail and will most probably create cache layers needed to not have to rebuild what is failing here) at https://app.circleci.com/pipelines/github/rbreslow/heads/4/workflows/8972250e-c1fb-4595-afdd-7f2c5d6e2b45/jobs/70?invite=true#step-106-635

Please simply cancel this build and restart it. Since musl-cross module definition is part of level 1 cache, it will invalidate hashes of level 2 and level 3 caches as described under #1288 (comment)

So no point into waiting that cache is being saved on this PR. Cancel that build and try again,