Since the fuzzers manage their own mongod topology, the current code that starts either a stand-alone or a replicated mongod instance needs to be extended to create disagg instances as well.

Once this is done, run the fuzzers and check for any BFs.

Risks for the project:

• that creating a disagg cluster requires a significant amount of code that somehow clashes with any current assumptions the fuzzers make that they run against either a stand-alone mongod instance or a replica dataset. There are several locations in the fuzzers that take different code paths for standalone vs. replicated setups.
• the new evergreen tasks start failing from day one, requiring extra work to stabilize. In the fuzzers, it is very difficult to skip a given test or an MQL construct, the test is expected to pass in its entirety.