`Sequence` cleanup: `fold` vs `reduce`, more powerful `scan`

[fridis]

foldf is redundant with reduce, and fold* generally provides the same functionality like reduce*, so we should think of better names, e.g.,

• reduce(R type, init R, f (R,T) -> R | abort R) R should be reduce_or_abort
• foldf(B type, f (B,T)->B, e B) should be reduce(B type, init B, f (B,T)->B)
• fold1(f (T,T)->T) should be reduce1(f (T,T)->T)
• fold(m Monoid T) should be reduce(m Monoid T)

Similarly, we need a similar set for functions for Sequence.scan.

[maxteufel]

I'm not sure about this, in most languages, fold seems to be the more general variant of reduce by taking an additional accumulator parameter.

We should certainly make this more consistent though.