Adapt system health monitoring section for 24h bandwidth aggregation period

We show aggregated bandwidth statistics over a 24h period on metrics.tpo, however claim

If you want to publish traffic statistics, you should aggregate all your relays' traffic over at least a week, then round that to the nearest 10 TiB (terabytes).

on the community portal (as noted in tpo/community/policies#28 (comment 3147888)). We should update the community portal text accordingly so it matches what we ourselves do on metrics.tpo (and thus implicitly recommend as a low bound of acceptable behavior).