Script/language-aware generic font-familiy resolving

Nobody has claimed this yet.

Currently `usvg` just naively iterates over all system fonts available and pick the first one that contains the expected character. Taking a tspan with a default generic family name `sans-serif` as an example:

```svg
你好
```

On my machine this makes usvg try to find a font that contains `你好`, ending up picking a random font `FZKTK--GBK1-0`, which it does contain the two Han characters but just, not expected. Expected way is to resolve to `Noto Sans CJK SC` or `Source Han Sans CN` or something more 'standard'.

System API integration is a way. On linux fontconfig is available. It has script-aware font matching scheme. Querying example:

```console
❯ fc-match ':charset=7424'
NotoSansCJK-Regular.ttc: "Noto Sans CJK SC" "Regular"
```

However system API integration may be bit heavy or dirty. Another way is, predefining font family names w.r.t. generic family names. As an example, here's the preferences file from Firefox: https://github.com/mozilla-firefox/firefox/blob/768652661f994d1bf533162e3c9ff8e0887d6784/modules/libpref/init/all.js#L2082-L2222.

Start by reviewing usvg's current generic-family font resolution and compare the script-aware charset matching described for fontconfig with Firefox's generic-family preferences. Done means generic families such as sans-serif choose an appropriate script-aware font for text like 你好 instead of an arbitrary font that merely contains the characters.

Written by the indexing model from the issue text.