Use sequence counters in cron helpers to limit execution time

[LegalizeAdulthood]

Even with cron scheduling the housekeeping tasks infrequently, with a large number of documents we can perform many, many thousands of HTTP HEAD requests. This results in a long bulk execution time for a cron job. A limit on cron job execution time should be set to perform scanning housekeeping in chunks.

To support incremental scanning, use a sequence id associated with each entity that results in a network request. Associate the sequence number with a timestamp. When an entity is scanned, it's sequence number is set to the current scan time stamp. When a scanning time limit has been exceeded, the cron helper will exit.

On the next scan, a new sequence number/timestamp pair is created. Scanning preference is given to to items with a lower sequence number. As chunks of items are scanned, the process will wrap around to the least recently scanned items.

Some care must be taken to bootstrap the sequence/timestamp housekeeping to prefer scanning items first that have no timestamp (e.g. have never been scanned) vs. those items that have been scanned but whose timestamp is the oldest.