Parallelize reputation_tracker and hivemind sync in CI

Run reputation_tracker in the background while hivemind syncs, then wait
for it to complete. This eliminates running reputation_tracker twice
(before and after hivemind) and reduces total CI sync time.

Both processes write to separate schemas (hivemind_app vs reptracker_app)
so there are no database conflicts. Hivemind uses LEFT JOIN with COALESCE
for reputation data, so missing data during sync is handled gracefully.