Consider switching the megazords to thinlto

[thomcc]

Surprised there's not an issue on file for this already. It should be a 1 line patch (change

application-services/Cargo.toml

Line 40 in 72bbb67

lto = true

to lto = "thin"), but requires testing on all supported platforms, to ensure we don't regress things WRT rust-lang/rust#50007 (It might be a good thing to do after we have semi-robust CI for them? Or could test manually).

Should speed up android builds a decent amount, among others.

┆Issue is synchronized with this Jira Task
┆Epic: important not urgent
┆Sprint: Backlog

[rfk]

(It might be a good thing to do after we have semi-robust CI for them? Or could test manually)

To ensure I understand what you mean here, since we have a KR about test coverage for Q1 - do you mean that our current tests run in CI are likely not sufficient to catch the kinds of missing-symbols errors that were seen with rust-lang/rust#50007?

[thomcc]

Right, since it causes a failure to load some (all) subparts of the megazord, but only on some OSes. I actually expect that our current android unit tests would catch it if it were obviously broken (e.g. if we deleted the lto=true line, or if thinlto didnt work for us at all), but given that its a platform dependent issue, it seems like we should check the platforms we care about