I was a life-long researcher who believes in the "null hypothesis" approach to making changes---a fancy way of saying one has actually to prove that making some change was beneficial. I have yet to see compelling evidence that going to a pure 64-bit moOde is beneficial.
But that's just me and my pessimistic nature.
Building a 64-bit kernel for the Raspberry Pi is pretty easy. The RPF provides nice tooling for cross-compiling on Intel/AMD Linux systems that makes the build process go fast (e.g., less than an hour on an old i7 box of mine).
I don't know what kernel bells and whistles you are alluding to. IMO the biggest challenge is proving all the audio-related kernel modules are 64-bit clean. Traditionally, latent coding errors crawl out of the woodwork when source code developed for a 32-bit system is recompiled to 64-bit and linked to 64-bit libraries. That's especially true for hardware drivers where often contract developers were involved.
Of course, moOde already offers a 64-bit kernel on an experimental basis and we continue to stumble over niggly problems.
Beyond the kernel, the entire moOde root file system has to be ported. As part of that, the moOde build recipe and the moOde builder scripts have to be adjusted to accommodate both 32-bit and 64-bit builds. This is because some RPi models contain ARMv6 CPUs and can't run a 64-bit OS. Until they are phased out, two moOde images and potentially two parallel support efforts will be required. This makes the effort look different to Tim and his merry band of helpers than it does to an individual.
'nuff said. Have fun, be sure to brew a large pot of coffee before you start, and don't forget to enjoy the music
Regards,
Kent