Need to be able to identify which SIDs / vbuckets are lagging for supportability

Currently when a customer issue is reported re: slow / stuck ingestion, the diagnostics available to identify which streams & vbuckets are at issue is limited. We should enhance our diagnostics to include.

1. per-sid seqno lags (currently we know which cids are lagging, but not which sid is in question. we night be able to correlate this today in KV logs, but in the future there will be a 1:N mapping of cids to sids, so we would need this anyway
2. per-VB seqno lags (currently we only know which cids are lagging- it could be all vbuckets or just one- we have no way of telling)

cc: Paolo Cocchi,Daniel Owen