examples/gkurve: improve unicode handling (fallback fonts, combining unicode characters, etc.)

[emidoots]

Right now after @PiergiorgioZagaria's work, gkurve tooling can render UTF32 fonts. What we would need to do to improve Unicode handling further is:

1. Add HarfBuzz bindings to our freetype bindings. I also heard MasterQ32 say he wants these.
2. Handle fallback fonts, e.g. try rendering with the desired font and then fallback to say Noto+Emoji font if the glyph is not found. In practice this means end user should provide a list of fonts, rather than just one, that we consume.
3. Verify we can render combining characters after hooking all this together. e.g. 👪 which is four UTF32 values.

Note many frameworks do not have this level of unicode support, e.g. Pygame doesn't either pygame/pygame#1436 - we should do better & have Unicode as a first-class concern.