Gonna rock the boat a bit here, but I don't think wanting to make a 2D game is a legitimate reason to not use DX12.
Sure, like I said originally, if your aims are academic or you're not at all concerned about having a very limited market of relatively-new, high-end GPUs on one specific OS, then knock yourself out and learn. But if you want to make money as an indie you want the broadest market available, and that can even include WinXP/D3D9 if you target the developing world (e.g. Korean internet cafes), there's just not good business to tie an otherwise simple game to rarified hardware requirements.
And there's no more sense (for a simple game, 2D or 3D) to target both new low-level APIs and the old traditional APIs -- either you are after total available market and old APIs cover it all, or you are after knowledge in which case the old API will hold you back from learning the new ones fully because of fundamental differences in threading approach (this effort/design tradeoff can be worthwhile for high-end games wanting best experience everywhere, however).
For me, I'm looking at D3D12 for Windows Store platforms (Win10, Phone, XBOne) and Vulkan elsewhere (including traditional Windows applications) which gives cross-platform and also some additional levels of hardware support and support for Windows 7. But I'm also very much informed by the fact that my aims will take years to accomplish and that this will be sideline income unless it turns out wildly successful.