@jmand Hey. Thanks for taking the time to test the nightly builds. It helps lots.
To be honest, we’re not having a great time since most of the developers in the team are super busy and have not been able to continue fixing bugs.
And as unfortunate as it sounds, fixing or implement other stuff has gotten more bugs 
I agree and have seen most of the things you’ve mentioned. The timeline keyframe creation has some sort of visual glitch that hides frames.
The tablet events that regulate selection of UI features went bonkers so selecting stuff is double difficult than it is to do with a mouse (particularly dragging the frames across the timeline…that makes me go crazy
)
I also reported one recently related to audio frames playback which crashed the app easily.
For example the copy / pasting got botched somehow and now it does not paste on the same place like before, which infuriates me because 95% of the time you only need to paste in the same place, not somewhere else, but this is a bug that got in accidentally like 1 year ago and I could only report it recently.
As for the layer tweening, yes I’ve asked that for years. Haven’t happened since there seems to be various prerequisites to make it work properly, which also depend on bug fixes and more implementation of base features that are missing (stuff you won’t see but that’s rather fundamental in the source code).
The camera thing was because someone wanted to do just that, and we weren’t going to say no. Sadly beggars can’t be choosers in the open-source world 
Also the camera was not actually complex because it wasn’t made from scratch, it’s piggy backing on the existing camera layer files and it had basic tweening capabilities since before we got here.
It just needed tweaking and we took advantage to improve on the UI and add extra features and whatnot, but if it was actually harder I assure you it probably wouldn’t have been implemented, ever.
As far as I understand, and while I’m probably missing something or could be mistaken, it’s not easy to propagate the changes from one layer type (camera) to the others since they are literally different structures that don’t have exactly the same properties. It’s like having 3 vehicles. They both have similar properties, but you can’t use a helicopter engine on a bike, it needs it’s own kind of engine, even if on the surface it’s considered the same thing.
For now you can try this: Animate your walk cycle centered in a different file. Then use the Camera motion tween features to move the walk cycle. Then export the animation as a PNG sequence and reimport it into a master Pencil2D file and keep working on the main file.
That’s what I do when I work with After Effects and other apps that do allow file link referencing.
As for the font tool. We know ppl want it. It’s been asked for years (not kidding) But it seems implementing such thing is super hard for this particular app (again as far as I understand since I’m not a dev).
So, until someone sits down to learn how it can be tackled, or gets interested enough to do it. I don’t see it happening soon
and to be honest I’d rather prefer having lasso selections first 
For now as a workaround what you can do is (and I know this works), have Pencil2D open at the same time as another app that does have a text tool. Something like Paint.Net or whatever.
Make sure you create the font over a transparent background and then copy and paste on a new bitmap layer in Pencil2D for quick work.
It will work easy as long as it’s bitmap text. Although vector shapes can also be copied and will become rasterized too 