split occurrence/inclusion-related operations to SearchableList

[gavinking]

Operations like firstOccurrence() and firstInclusion() suffered from several problems:

• since they were declared by the covariant type List, they weren't typesafe,
• they were required to be implemented by every List, even though they weren't really appropriate for e.g. LinkedList
• the implementations on String were inefficient
• the actual usecases for these operations called for additional optional parameters, e.g. from indexes

Therefore, they have been factored out to a new invariant interface SearchableList which is implemented by String, Array, and ArrayList. Since all these are invariant types, we've regained typesafety. In addition, optional parameters from and length have been added, and the implementations on String have been made much more efficient.

I have one more small remaining task here which is to decide between:

• defining all the operations in terms of occurrences() and inclusions()—making SearchableList easier to implement, or
• leaving it as it is, where occurrences() and inclusions() are defined in terms of the other operations of SearchableList—potentially more efficient.

[gavinking]

I went with the simpler implementation. I doubt it is that expensive to instantiate a comprehension.

This is now done.

[gavinking]

After benchmarking, I went back to something more like what I had before.