Enable fat link-time optimizations in release builds

[charlespierce]

Info

• Despite putting a large focus on performance (and to some extent, binary size), we haven't taken advantage of Rust's ability to enable link-time optimizations (LTO).
• These optimizations can improve performance and reduce binary size, at the cost of taking longer at compilation.

Changes

• Updated Cargo.toml to enable fat LTO in release builds of Volta
• Also added codegen-units = 1, which is another setting which increases final performance at the cost of compile time (see the Rust Performance Book for more details).

Tested

• On Windows, this results in an ~10% reduction in binary size. I didn't see any significant performance difference.
• On macOS (M2 Max), I see ~25% reduction in binary size (!) and ~5% faster performance for a simple test.
• The CI builds take slightly longer, but not massively so to the point where running them with full optimizations would impact PRs.