Fix data on objc_library

[keith]

Previously this was just ignored, now it is correctly propagated.

Fixes bazelbuild/bazel#16270

[pzembrod]

This breaks a number of internal targets; most with a 'UIKit/UIKit.h' file not found error.

Maybe most targeted is a failure of https://github.com/bazelbuild/bazel/blob/master/src/test/java/com/google/devtools/build/lib/rules/objc/ObjcLibraryTest.java:

ObjcLibraryTest.testArtifactsToAlwaysBuild fails:

value of: sortedBaseNamesOf(...)
expected: x.o x2.o y.o y2.o
but was : libx.a liby.a x.o x2.o y.o y2.o
	at com.google.devtools.build.lib.rules.objc.ObjcLibraryTest.testArtifactsToAlwaysBuild([ObjcLibraryTest.java:1909](https://cs.corp.google.com/#search/&q=f:com/google/devtools/build/lib/rules/objc/ObjcLibraryTest.java:1909&ws=tap-presubmit-rerun-server/134289872&snapshot=2))

Does that repro when running the test with Bazel?

[pzembrod]

Fwiw, almost all tests pass for me if I just comment out the default_runfiles = runfiles, param in the DefaultInfo constructor.

And on that note: would following the second "avoid" in https://bazel.build/extending/rules#runfiles_features_to_avoid be useful guidance here? Honest question on my side, as I don't (yet) know much about this "avoid" guidance.

[keith]

And on that note: would following the second "avoid" in bazel.build/extending/rules#runfiles_features_to_avoid be useful guidance here? Honest question on my side, as I don't (yet) know much about this "avoid" guidance.

It's a good question on this weirdo branch specifically, especially since we're adding something new. I copied this from cc_library since semantically I think they should work the same way. but idk how common the "old way" is

[keith]

I fixed the issue with the .a being in the runfiles, that was a bug, updated the tests to verify it as well