fix a flake in the sync manager edge case test

[Stebalien]

Depending on scheduling, we may see c2 before we see c1.

[schomatis]

LGTM I can't reproduce locally (after 1000's of runs) but this seems like it should fix the issue.

I'm deferring to @vyzo for the actual approval; accepting either c2 or c1 seems to defeat the purpose of the test altogether. I'm not sure how the done() API works but, as Sync() can in theory be run in parallel, to process tipsets in the proper order I suppose that this test is working under the assumption that syncManager is going to throttle them accordingly (coming from the same parent). If this is not the case then I have no idea what this test is doing and my green tick shouldn't count anyways.

[Stebalien]

I believe the goal is that c1 should be the final decision, even if c1 comes first. This still tests that (ish).

[schomatis]

Looking at this again, c1 and c2 do not come from the same parent as I originally thought. Sync() calls will be issued in parallel for both, so this is in fact racy (not sure if by design). The fix makes sense as a temporary solution.

[vyzo]

yes, that's the intention! thanks for fixing this.

[magik6k]

❤️