refactor(swingset-liveslots): rep vs instance label tests more similar

[erights]

Improve the parallelism of heap instance labeling test vs representative labeling test. In particular, add a passStyleOf tests that had failed before #7273

[erights]

I wrote the following text before #7273 and am preserving it as a historical note.

After #7410 was merged, I suddenly wondered why @warner 's added test of representative labeling seemed to work prior to the endo reconciliation. By contrast, I had to suppress the similar tests in test-label-instance.js with a test.skip, because until the reconciliation, the new @@toStringTag data property violates the remotable invariants. Thus, prior to the reconciliation, the labeled instances or representatives are not even Passable. passStyleOf on them throws.

The reason @warner 's test passes anyway is that it doesn't do a passStyleOf, or indeed anything that is actually sensitive to whether these allegedly virtual objects are actually Passable. Frankly, I'm surprised that it was able to get that far. (@warner @FUDCo , Perhaps that indicates we need an additional validation test somewhere?)

In any case, this PR adds the passStyleOf tests that should be there, restoring the parallelism with the instance labeling test. Doing so did cause the expected failure. Good! So the price is that we needed to mark these as test.skip, just as we had to do with the instance labeling tests, with a TODO to unskip following the reconciliation.

[warner]

LGTM