Run slow tests in parallel to improve overall test time

[IamTheFij]

This is one mitigation for #935 and cuts test runtime in half for me.

Note regarding the change to the expected results for the reverse sort: I'm not sure exactly why it was passing before. It's possible that the slice was modified by one of the earlier tests requiring the result to be not quite the proper reversed version. This patch makes sure the slices aren't modified and also fixes the expected result.

[IamTheFij]

It's interesting that the tests when run on GitHub show next to no difference. This appears to be because the tests that rely on HTTP requests already run quickly here compared to when run on a local machine.

#19 106.9 --- PASS: TestCompareDigest (0.24s)
#19 106.9 === CONT  TestManifestMultiUpdatedPlatform
#19 106.9 --- PASS: TestManifest (0.29s)
#19 106.9 --- PASS: TestManifestVariant (0.65s)
#19 106.9 --- PASS: TestManifestMultiNotUpdatedPlatform (0.74s)
#19 106.9 --- PASS: TestManifestMultiUpdatedPlatform (0.60s)

Running those in parallel saved about 1.5 seconds here vs 30s on my local machine.

[crazy-max]

Small nit otherwise LGTM!