Simplify tests with runSkycfgTests test runner

[seena-stripe]

Summary

Adds a test runner runSkycfgTests to remove redundant testing logic. Many of the tests had a 1 line skycfg expression and a 1 line desired output, but duplicated all the test boilerplate, so this PR adds a test runner to simplify setting up a new test

This is purely cleanup and shouldn't affect evaluation.

Example

 func TestUnsetProto2Fields(t *testing.T) {
        // Proto v2 distinguishes between unset and set-to-empty.
-       msg, err := eval(`proto.package("skycfg.test_proto").MessageV2(
-                f_string = None,
-        )`, nil)
-       if err != nil {
-               t.Fatal(err)
-       }
-       got := mustProtoMessage(t, msg)
-       want := &pb.MessageV2{
-               FString: nil,
-       }
-       checkProtoEqual(t, want, got)
+       runSkycfgTests(t, []skycfgTest{
+               {
+                       src: `proto.package("skycfg.test_proto").MessageV2(
+                               f_string = None,
+                       )`,
+                       want: &pb.MessageV2{
+                               FString: nil,
+                       },
+               },
+       })
 }

This doesn't save on LoC very much but reduces the boilerplate so the test code is a lot more signalful and test output is consistent

[kathleen-stripe]

Only the want != nil && got == nil change is blocking. Thanks for doing this refactor!