docs: elaborate on event deletion/retention

[leviramsey]

There are some counter-intuitive aspects around event retention and deletion, especially if the aim is to limit disk consumption in a Cassandra cluster. This change is intended to elaborate on some implications of this.

Explicitly not covered:

• Cassandra cluster sizing based on retention period and write rate, as there are more considerations at play
• What sort of operational tasks become required when deleting events with less than 50% free disk space on a Cassandra node, as this will be very idiosyncratic
• Using a different compaction strategy: the LeveledCompactionStrategy may be worth considering for Cassandra clusters which can't easily scale out, at the expense of write throughput, but discussion of that is out of scope for these docs

[patriknw]

LGTM, thanks