Benchmark 1,2,3-dimensional functions against general n-dim ones

[egpbos]

We should benchmark whether the 1-3 dimensional special functions are actually faster than the n-dimensional one. If not, we can just use one implementation function for all ranks, with simple wrappers for the 1, 2 and 3 dimensional cases. This would reduce the number of implemented functions from 72 to 18, plus 54 simple wrapper functions.

[egpbos]

See commit message above; the general n-dimensional functions in the current simple numpy-like interface are exactly the same! Will have to test more with more high-performance planner flags, though (#7).