better text measurement plus UTF-8 support

[nptr]

Hello Werner,

of recently I had the need for freetype font rendering but my needs exceed the current capabilities it seems. I took the freedom to expand the library a bit and would like to propose this changes:

• UTF-8 support for text rendering and measurement where possible (freetype)
• More accurate text measurement that uses a per-glyph-width instead of fixed-width.
• Introduce bm_text_measure() to measure text width and height in one pass where possible. (Maybe replacing bm_text_width() and bm_text_height() some time?)

Remarks

• The UTF-8 support I added is nice, CJK looks reasonable for example. But it won't render complex scripts like arabic. This would require an additional pass with a text-shaping library and is out of scope.
• Text measurement is tailored for horizontal scripts. It assumes dynamic glyph width, but a fixed height. There might be vertical scripts that need it the other way around, I dont know. Also out-of-scope I'd say.

ToDos

• [x] Checking-in Arial Unicode with 22 megs is not exactly nice. I should find a solution for that.
• [x] I removed \t handling for text measurement. I should add it again.
• [x] _utf8_get_next_codepoint() is duplicated (bmp.c and ftfont.c) because I didn't want to expose it in bmp.h.

Please let me know what you think, what works and what doesn't work for you. I'll try to accommodate the feedback in follow-up commits and then squash them into one.

[wernsey]

This is awesome, thanks.

I'm from an English-speaking country so I don't have a lot of experience with UTF-8 and I18n. These are desirable features, though, so I really appreciate you taking the time implementing them.