perf(tests): Enable optimizations for benchmarks

[ghost]

Closes #2326.
Closes #1798.

[LucioFranco]

do you have any bench numbers for this?

[binarylogic]

I've updated the issue to close #1798 as well.

[Hoverbear]

Luke noted in #2326 (comment) debug is on intentionally for profiling reasons.

[lukesteensen]

The defaults are here:

[profile.bench]
opt-level = 3
debug = false
debug-assertions = false
overflow-checks = false
lto = false
panic = 'unwind'    # This setting is always ignored.
incremental = false
codegen-units = 16
rpath = false

So the diff here is really lto and codegen-units (I'd like to keep debug = true so we can still profile benchmarks by default). I imagine the biggest difference here is in build times. I worry that significantly increasing those will slow down our iteration time on benches, but it would be nice to have those numbers be better representative of production builds.

Do you have any numbers on the difference in build time for benchmarks with the above changes?

[binarylogic]

@a-rodin is this still relevant given the comments?

[binarylogic]

@a-rodin mind chiming in here?

[binarylogic]

@Hoverbear should we close this? If so, can we address the linked issues as well?

[Hoverbear]

CLosing...