Cool article. Here are my thoughts for ROM hacking only:
Pros:
I think the biggest advantage of a GBA Game/ROM Hack is that it's automatically a cross platform game that you can also play on mobile. The 60fps is a nice tack on because the GBA's refresh rate is locked on ~60 fps.
I'm not sure if the tools sentiment applies to Gen II or Gen I hacking which are already using the Disms, in gen III however, A-map is probably the only problem child. XSE does it's job in compiling scripts, and most misc tasks we have commandline tools for automation.
The barrier to entry in making a basic ROM Hack is pretty much zero though. There exists ROM base patches or dynamic engine resources which will update your ROM to near latest gen with a lot of cool features in a few clicks. You *could* make a game without seeing a single line of C or ASM, as our predecessors did in a lot of the 2014 HOTY and earlier hacks. Overall not having to make a game engine or write any custom code is probably a reason ROM hacking Gen III boomed around 2008.
ROM hacks are also really popular and there are plenty of hacks to extract example implementations from. Most of our customized code is open source and available for modification/download for those who wish to do that.
Cons:
You do feel the limitations on the GBA when working with the smaller screen resolution and limited color depth. It's not exactly the fastest engine either, and emulators are often inconsistent with code implementations working on some emulators but not on others. Binary ROM development debugging tools are also pretty limited, and A-map can turn sour if you look at it the wrong way.
As for writing custom code, using GF's functions abstract a lot of the math and hardware implications, but its still necessary. IO mapped registers can be a pain sometimes, and a lot of cool tricks aren't documented anywhere (128x128 OAMs, Hardware Blending single sprite when multiple share the same obj window/priority ect.)