Add support for ot-metrics

[skef]

Some test infrastructure I'm fooling with would benefit from this.

Hoping you might have some advice on whether or how I should test the x/y_variation functions, or if I should just delete those ...

[anthrotype]

thanks!

some advice on whether or how I should test the x/y_variation functions, or if I should just delete those ...

what's different/special about those as far as testing is concerned?

[skef]

what's different/special about those as far as testing is concerned?

Looking at the source, they seem to scale the returned values based on em_scalef_x or em_scalef_y, which use scaling factors in the font. But I didn't dig too deeply on how to set those factors. @khaledhosny might have a better idea.

[skef]

After thinking about this I added a mundane test for each of the methods. Given that uharfbuzz is a very thin layer over HarfBuzz, that's probably fine. Even if these methods aren't useful now, perhaps future additions will make them so.

[khaledhosny]

I’m not familiar with this API, to be honest.

[anthrotype]

fwiw i'd be fine with merging this as is

[skef]

If it passes review, including verifying that I'm calling each method I added in the tests in a somewhat substantive way, I think it will be fine.

[skef]

Sorry, don't know where my head was on this one (on the typing that is -- with the variable stuff I was mostly copying from other parts of the file, unsure of how reliable the cython conversion would be).