Clojurians Log v2

I am working on distributed GC support now, if you have special requirements lmk. I have been using BranchBench, https://arxiv.org/abs/2604.17180, to measure Datahike's performance. It is pretty good in general, but GC was still a weak spot. We will durably cache marked elements to speed up GC iterations.

Heads up. After distributed incremental GC is done, I would like to do one index format change. While working towards completion on pg-datahike it turned out that our comparator sorts NaNs differently. Fixing it will require a breaking change to the index layout. I have been saturating Postgres tests for this very purpose before hitting the final work towards 1.0. In this step I would also replace the Fressian serializer with https://github.com/replikativ/boring (CBOR). There is also an edge case bug for the datoms iterator on upserts where it misses some elements that will be fixed with this. The migration code should make this not too painful to do now, but in general I have tried to avoid this as much as possible over the years as I know the downstream work it causes. Lmk what you all think and what you need.