Un-xfail test_compare_against_quark

[renxida]

Had to xfail this test due to a very minor over-tolerance during IREE bump #1003

It doesn't look like much of a difference and only barely exceeds the tolerances. However, I don't know this test well enough to go in and increase the tolerances.

---

Failure log: https://gist.github.com/renxida/f1af75da6d9a82e662c6824fa1314ec7

Error message:
FAILED sharktank/tests/models/llama/quark_parity_test.py::QuarkParityTest::test_compare_against_quark - AssertionError: Tensor-likes are not equal!

Details:
Mismatched elements: 1 / 16384 (0.0%)
Greatest absolute difference: 0.00390625 at index (0, 1, 713)
Greatest relative difference: 0.004730224609375 at index (0, 1, 713)

---

Will assign to test author after #1003 merges to take a look at this and maybe increase the tolerance and re-enable if that's appropriate.