Fully optimize cargo release builds and add debuggable release profile

[SUPERCILEX]

See #595 (comment)

[SUPERCILEX]

Ah, looks like the tests are using rust 1.52 but this needs 1.57: rust-lang/cargo#9943

[SUPERCILEX]

@BurntSushi is there any interest in this? I'm wondering if it's worth bumping the MSRV.

[BurntSushi]

I'm totally fine to bump the MSRV, up to and including the latest Rust stable release.

In general, I don't really look at PRs like this unfortunately until I actually do a ripgrep release. At which point, I'll decide whether to bring it in by actually trying it out and playing around with it. If changes need to be made, I'll just do them by squashing them into your commit. If I didn't do that, then I'd be context switching all of the time and would get nothing done. Since I became a Dad, my release cycles have become longer.

So this PR might sit for some time. But I might be interested. Hard to say without actually giving it a whirl and seeing what it's actually like.

[SUPERCILEX]

Ok, sounds good! I bumped the MSRV so this should be ready to merge if you decide you like it.

BTW, dunno if people have mentioned this before, but a big reason for supporting cargo installation is that you can use rustflags = ["-C", "target-cpu=native"] which will presumably get you slightly higher performance.

[BurntSushi]

BTW, dunno if people have mentioned this before, but a big reason for supporting cargo installation is that you can use rustflags = ["-C", "target-cpu=native"] which will presumably get you slightly higher performance.

Yeah I'm aware of that. I personally think cargo install is just more about convenience. Setting target-cpu=native is likely to help a little bit in certain workloads, but ripgrep already uses SIMD in its hot paths even when target-cpu is not set at all. I don't even bother compiling with target-cpu=native when benchmarking ripgrep. (Partially because I've never really seen it make much of a difference and partially because it's an uncommon build configuration.)