Adjust cleanAllRUV max per txn and interval limits

[mreynolds389]

Issue Description

With very large replication change logs the cleanAllRUV task change log purging consumes a lot of CPU and can render the server unresponsive.

The cleanAllRUV changelog purging and the "normal" changelog trimming are using different limits. The cleanAllRUV "purging" holds the transaction 10 times longer than the "trimming" does. When we use the same/smaller limits the CPU still gets high but it's spread out and it should not block other threads from running.

[tbordaz]

My understanding is that changelog trimming use maxage or maxentries. So when the thread hit one of those limit, it can stop. For example if CL contains 1M entries from the past week and maxage=7days, trimming can stop without checking the most recent 1M entries.
With changelog purging, there is not such limit. The purging thread has to go through each entries to purge the update or not. With the same example, the purging can not stop and need to evaluation the most recent 1M entries.

Can this difference be responsible of the difference of duration "trimming" vs "purging" ?

[mreynolds389]

97e9106..d89715d 389-ds-base-3.0 -> 389-ds-base-3.0
8d2300a..eba9079 389-ds-base-2.5 -> 389-ds-base-2.5
24e5980..29de3fa 389-ds-base-2.4 -> 389-ds-base-2.4
dd9d04a..ce0196d 389-ds-base-2.3 -> 389-ds-base-2.3