Gradual attribute changes

[daveyarwood]

Moved from #179.

Quoting myself:

Currently, translating things like crescendos (gradually increase volume) and ritardandos (gradually decrease tempo) is cumbersome, as you have to manually adjust the attribute over time, like this:

(vol 50) c8 (vol 60) d (vol 70) e (vol 80) f (vol 90) g2

It would be nice if we had a way to gradually apply an attribute over an established period of time, maybe something like:

(vol>) c8 d e f (<vol 50) g2

[daveyarwood]

I had a related idea for "automation voices": https://github.com/alda-lang/alda-core/issues/80#issuecomment-712809114

I still like the syntax idea in my comment above because sometimes, it might be easier/more desirable to write the attribute values inline instead of as a separate "voice". I think we could add both features.