I think a lot of these Bitcoin developers are bored with the project... because it is mature and secure and there are not a lot of challenges to keep them busy.
it seems to me that there are a few challenges yet
- more network partition resistance
- obfuscating the bitcoin protocol over the public internet
- the eventual direction for additional throughput scaling
all are in the works, but they're long term projects that aren't really gradual changes. The mailing list is still busy, so I think it's safe to say that these big projects will both continue and proliferate.
but as far as the big job of getting the codebase more sensibly structured, alot of the work is now done, and many parts might now stay the same long into the future. you can see the results of that with the so-called kernel project; it probably wouldn't have been possible to start the kernel library (which is just the consensus logic factored out into a standalone library) if the codebase wasn't slowly tidied up and consolidated over time.
Marco's job was to oversee alot of that churning consolidation over the year as an expert on the c++ language. I can see how he might find now a good time to move on from maintaining, as there will be less need for that kind of role which seemed to suit him so well. he's still active on the project though, so this is more what you could call "shuffling" roles instead of "moving on" outright