Add linux-musl entry for FALLBACK_HOST_RID

[am11]

Fixes #65152

[ghost]

Tagging subscribers to this area: @vitek-karas, @agocke, @VSadov
See info in area-owners.md if you want to be subscribed.

Issue Details

---

Fixes #65152

Author:	am11
Assignees:	-
Labels:	area-AssemblyLoader-coreclr
Milestone:	-

[am11]

cc @janvorli, @vitek-karas, I have used same detection mechanism as we have used in:

runtime/eng/native/init-distro-rid.sh

Lines 163 to 164 in 8e4bef2

	if "${rootfsDir}/usr/bin/ldd" --version 2>&1 | grep -q musl ||
	strings "${rootfsDir}/usr/bin/ldd" 2>&1 | grep -q musl; then

which was also ported to https://github.com/dotnet/install-scripts/blob/11b4eebe23d871c074364940d301c3eb53e7c7ec/src/dotnet-install.sh#L188

as fragile as it seems, it just works.

[vitek-karas]

Looks good, but I definitely no expert in the CMake magic - so let's wait on @janvorli to take a look as well.

/cc @VSadov

[vitek-karas]

Actually - spoke too soon. The version-less windows RID breaks some tests. I looked at one of them (PortableComponentOnSelfContainedAppRidAssetResolution) and it's basically the test depending on the specific RID we fallback to. It should be pretty straightforward to modify the test to expect the new behavior (or change the setup of the test to accommodate the change).

That said - maybe we should split these changes into two different PRs.

[janvorli]

The change looks good, but I leave the fallback rid resolution decision to the host folks.

[vitek-karas]

@am11 could you please remove the win RID change and let's merge the linux-musl part on its own.
I'm fine with the win change as well, but it will need some test tweaks, so let's not block on that for now.

[am11]

@vitek-karas, I have removed the win commit for later time.

BTW, looks like we need to unescape linefeeds in failing test logs:

[vitek-karas]

I checked offline with others that all the failures in the CI run are unrelated (and mostly known problems which are being worked on).

[vitek-karas]

Re escaping newlines - I haven't looked into that yet, but I suspect it's some weird interaction with the testing framework.

[vitek-karas]

Thanks a lot @am11!