rndc dumpdb -expired doesn't always work

Per Support ticket #20909, I asked for a cache dump that included expired content (lots, as indicated by the stats) but got a dump that included only active content.

This new feature was introduced in #1870 (closed)

I strongly suspect that it doesn't output expired content unless we have stale cache enabled. In this instance stale cache was disabled, but it looked as if the cache was full of expired content and very little active content (per the stats output). I was very very disappointed when I received the cache dump, and then further checked (from the logs) that the 'expired' option had been included when it was requested.

Please fix!

Also I have a secondary question, inspired by the original issue #1870 (closed), and following discussion with BIND engineers.

As well as expired content, cache may also contain replaced content - that is, older versions of RRsets that have since been updated. I had this highlighted to me when discussing with Engineering the dumpdb output (and the problem I am trying to troubleshoot that I hoped this dumpdb would help with). This is the question that was posed in the original issue during review:

There may also be cases where a single node has multiple rdatasets/rdataslabs associated with it for the same RRtype - only one of which is "current". These expired and not-yet reclaimed data should be dumped as well.

I don't know that this is a problem, but I don't know that it isn't either.

Do we also dump replaced content as well as expired content with the -expired dumpdb option? (This might required a separate and new feature request if it doesn't).