Added nth to BitVec::Iter so that .skip(n) gets optimized

[hacatu]

Fixes #122

[pczarn]

We could have it documented that nth/skip is O(1) so nobody wonders about the reason this code is necessary (which is time complexity)

[pczarn]

Could use a benchmark as well

[hacatu]

Ok, I added a benchmark, basically create a bitvec where we skip the first 3 << 20 entries and read the last 16411 (much smaller) entries within the benchmark. This confirms what we expect:
before:

1,153,510.30 ns/iter (+/- 2,229.89)

after:

6,091.13 ns/iter (+/- 142.41)

I added an explanation that Iterator::skip will use Iterator::nth to the nth override, let me know if I should expand it

[pczarn]

Thanks a lot!