test: Use `set sysroot` for more NDK compatibility

[alexcrichton]

Recent versions of the Android NDK no longer ship debuggers like
arm-linux-androideabi-gdb, but instead one prebuilt binary gdb. We can
symlink this into place at least to get our detection still working, but it now
needs to be told what the sysroot is so it can correctly do... something. Long
story short, tests didn't pass with this change and after this change they pass.

[rust-highfive]

r? @brson

(rust_highfive has picked a reviewer for you, use r? to override)

[alexcrichton]

r? @michaelwoerister

This, in combination with #33832 should enable us to move to the r11c NDK and API level 9 for our Android builders

[michaelwoerister]

@bors: r+

[bors]

📌 Commit 0f3e865 has been approved by michaelwoerister

[alexcrichton]

Ok, I've now verified that this plus #33832 gets a green build on Android, bringing improvements like:

• The ARM and x86 API level for nightlies goes from 18 to 9
• An AArch64 nightly is also created
• We're using the r11c NDK (most recent)

I haven't deployed the buildbot changes just yet, and I'm hesitant to do so. These two changes are required and not present on the beta branch, so if we upgrade buildbot we'll be unable to build a new beta compiler (unless we use a different image to build a beta compiler than the nightly compiler. To me, there's a few courses of action to take here:

1. We wait for these changes to make their way to the beta channel, then we upgrade the bots. This means that all channels can build on the new bots, but delays these changes by another 6 weeks (notably releasing the aarch64-linux-android target).
2. We backport these two changes to beta and immediately upgrade the bots. We'll be in trouble if we need a stable point release, but we can make it work if we really need to do that.
3. We add separate bots for the beta channel for the time being and upgrade the nightly-only bots.

My preferred course of action is (2) as it's the easiest and probably doesn't have any impact anyway. This change isn't related to the build at all and #33832 should in theory be pretty minor as well. As a result I'm going to beta-nominate this and that, and we can discuss.

[alexcrichton]

(for those cc'd, see my previous comment)

cc @froydnj - going back to API level 9
cc @pietro - adding aarch64 android

[froydnj]

We backport these two changes to beta and immediately upgrade the bots. We'll be in trouble if we need a stable point release, but we can make it work if we really need to do that.

This seems like a reasonable course of action for propagating Android changes sooner.

[michaelwoerister]

Given the minor impact of the changes, backporting seems fine to me.

[Nercury]

In this case, we should also get libc tests going on Android's AArch64.

[bors]

⌛️ Testing commit 0f3e865 with merge 397cfae...

[bors]

☀️ Test successful - auto-linux-32-nopt-t, auto-linux-32-opt, auto-linux-32cross-opt, auto-linux-64-cargotest, auto-linux-64-cross-armhf, auto-linux-64-cross-armsf, auto-linux-64-cross-freebsd, auto-linux-64-cross-netbsd, auto-linux-64-debug-opt, auto-linux-64-nopt-t, auto-linux-64-opt, auto-linux-64-opt-mir, auto-linux-64-opt-rustbuild, auto-linux-64-x-android-t, auto-linux-cross-opt, auto-linux-musl-64-opt, auto-mac-32-opt, auto-mac-64-nopt-t, auto-mac-64-opt, auto-mac-64-opt-rustbuild, auto-mac-cross-ios-opt, auto-win-gnu-32-nopt-t, auto-win-gnu-32-opt, auto-win-gnu-32-opt-rustbuild, auto-win-gnu-64-nopt-t, auto-win-gnu-64-opt, auto-win-msvc-32-cross-opt, auto-win-msvc-32-opt, auto-win-msvc-64-cargotest, auto-win-msvc-64-opt, auto-win-msvc-64-opt-mir, auto-win-msvc-64-opt-rustbuild

[alexcrichton]

Discussed with @brson today and like with #33832 we concluded this would be good for backporting.