This criticism doesn't make a ton of sense from a practical development standpoint in modern times, people continue to rehash the same OpenGL vs DX argument from the 90s and apply it to Metal vs Vulkan. Now, almost everyone that isn't an engine author or a large AAA game developer is using a game engine (Unity, Unreal, Game Maker, etc) and the few developers that don't are building graphically simple games where OpenGL is completely sufficient or have technical ability to port to a few different APIs without issues.
There are a couple of games with Vulkan renderers that don't support Metal that I can see this being an issue in (Doom being a good example), but id is still a AAA studio that could have someone write a Metal backend if Mac made sense to them (same way they have DX12 and gnm backends for ps4 and xbox one).
Another thing is people often assume the only cost involved in porting a game to a new platform is graphics API, which is strange, but makes sense if your exposure to graphics programming is NeHe tutorials or something. All modern engines use thin wrappers over the system graphics API and low level code that directly calls OpenGL, DX, Vulkan, Metal, etc is a very small percentage of total code. Rants about graphics API being the main bottleneck to game portability annoy me a little bit also because they ignore portability challenges that are not graphics related, and even graphics portability challenges that are not directly related to API. It's naive to think if everyone just supported Vulkan all games could be instantly compiled to every platform or even that low level graphics API calls are the hardest thing to port. I agree that metal makes things a little harder by introducing a new shading language, but cross compiling to other shader languages from hlsl or glsl is standard and mostly a solved problem by now. If you're a small developer that doesn't or can't write a thin low level graphics wrapper for your game there are good opensource libraries that will do this for you (bgfx, oryol).