There was some discussion on this a while back and basically my recollection of the outcome was that the benefit of migrating to a different RDBMS was not really worth the effort.
The database is primarily used for configuration storage, there is not really much of a transactional need.
In terms of DBA'ing multiple types of databases, there is little to no maintaintence and administration performed on the databases, since they are not used transactionally and do not see large growth figures or cause any performance bottlenecks.
My feeling is that there is enough work still to do in getting the software to work faultlessly and adding new technologies and devices that adding a new variable such as multiple DB support creates unneccessary work.
But that said, if you are keen to experiment - it would likely give you quite a good exposure to the database dependencies and that information would be useful to the rest of the community documented in the wiki, even if the port proved fruitless.
At a guess I would expect most of the SQL dispersed throughout the system to be quite portable and if memory serves me correctly I think that the first version of Pluto was actually written to use MS SQL Server - but I could be wrong.
Your primary DB dependencies will be found in the Object-DB wrapper that is written in C++, the many shell scripts for setup and bootup checks and as you mentioned the web admin server code (which is mainly php).
I would suspect that there is an element of automatic timestamping of rows utilised, especially by the sqlCVS stuff - not sure how well this would port.