Fix our xz2 crate configuration to enable static linking.

[jsirois]

Previously liblzma was dynamically linked (to a homebrew version).

Fixes #172

[jsirois]

The linking is now:

$ otool -L dist/scie-jump-macos-aarch64
dist/scie-jump-macos-aarch64:
	/usr/lib/libiconv.2.dylib (compatibility version 7.0.0, current version 7.0.0)
	/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 1336.0.0)

[jsirois]

The fix is good, but I'm in a weird state where cargo +nightly fmt is giving different results on the Linux and Mac runners.

[huonw]

I'm assuming that the Cargo.toml change is the only real change, and the code changes are "just" reformatting?

Are you aware that rustup supports pinning to particular nightlies via nightly-$date? For instance, rustup toolchain add nightly-2023-11-20. https://rust-lang.github.io/rustup/concepts/toolchains.html#toolchain-specification This might allow getting stable formatting that only changes when explicitly desired, or at least pinning to a version that doesn't have the inconsistencies.

[jsirois]

Are you aware that rustup supports pinning to particular nightlies via nightly-$date

Yeah, but I've never had to do this until now - totally unclear why. Even having to use nightly at all is horrible. I'm not sure what is up with the glacial pace of the promotion of rustfmt nightly features compared to rust otherwise.

[jsirois]

This is a very crappy situation. Wanting to use stable, and encoding that in rust-toolchain so all devs get things working, but then using nightly for rustfmt. Afaict I'm forced to add docs or an outer script or both to make sure fmt just works as well. Having to use a magic nightly pinned version is going to skate right by a new dev. In fact, it will mess me up too I'm sure when I forget a month from now.

[jsirois]

I'm going to sink some time into this rustfmt reproducibility situation before landing.

[jsirois]

Alright, much headbanging but I should have looked at issues: rust-lang/rustfmt#5964

Still not sure what to do, but this appears to be a ~temporary problem.

[jsirois]

Alright - the rustfmt noise is isolated to some ~commented out CI steps and tracked in #174.

I'll merge once green since those CI fixes are minimal despite being orthogonal.