Based on your patch level, my guess is you’ve been bitten by the Daylight Saving Time bug which causes an extremely large number of failed instances for daily scheduled jobs. Search for “DST” or “daylight” to find plenty of threads discussing this issue.
You can run the following query in Query Builder to perform a quick check and verify this is indeed the problem:
select count(si_id) from ci_infoobjects where si_schedule_status = 3
A large number (e.g. > 100,000) of failed instances likely indicates the DST bug is the issue. Fair warning: the query could take a while to run and may timeout.
4 failed instances certainly isn’t enough to cause the CMS table to explode in size. I am not sure what else would cause this to occur. I recommend opening a ticket with SAP especially if the table is still growing at a rapid pace. They may have a tool to identify what went awry. There is also a repository documentor in the downloads section which may prove useful to find the rogue objects.