This is a relatively complicated matter, due to the license minefields in combination with Oracle policies. Let's start with the biggest blockers to the less important ones:
- The VNC support depends on libvnc, which is covered by GPLv2 (and there seems to be no alternative), and we can't get legal approval to add a 3rd party library dependency with this license to any part of VirtualBox, be it the base package or the Oracle extension pack, period. LGPLv2 is the limit, and arguing doesn't help. We tried and failed. So this is the reason why I can't see this happen ever.
- The VNC support works OK, but has in comparison to VRDP very limited functionality. In addition the code is not actively maintained, the last change to it was over a year ago, adding basic IPv6 support. It's a 3rd party contribution which we added to the tree so that it's easy for someone not affected by item 1 to build/publish an extension pack. We will not invest time into the VNC code beyond keeping it alive. Anyone interested in improvements: feel free to contribute, we'll gladly accept patches.
- Why would anyone want to use VNC if there is a better option available? VNC loses against RDP (and VRDP) in almost any criteria, and from my experience there are excellent RDP clients available for any relevant client platform.
- I see no convincing use case for having some VMs use VRDP and some VNC, even though that's possible with the current config options.
Bottom line: we simply can't do what you're dreaming of, and the only way to get out of this stalemate is someone else publishing the VNC extension pack. This was the motivation for pushing the contributors to redo the initial VNC implementation (integrated mid 2010 as an ugly hack to VBoxHeadless) as an extension pack using the same new VRDE infrastructure as VRDP is using (happend in early 2012). Since then it's feasible for 3rd parties to provide binaries so that not everyone needs to rebuild.