mixing weighted and unweighted iterators in normalizers

[Peter]

Currently (0.5) the operator() in range normalizers have an restriction that input and output must either both be weighted or both be unweighted. Is there a good reason for this? For instance, in qQuantileNormalizer, we have:

template<typename RandomAccessIterator1, typename RandomAccessIterator2>
void operator()(RandomAccessIterator1 first, RandomAccessIterator1 last,
                RandomAccessIterator2 result) const;

and the docs says: RandomAccessIterator1's value type is convertible to RandomAccessIterator2's value type

I wonder if this is really needed, in other words, the weights are just copied (not modified) so it should be possible to modify the algorithm to allow us to mix weighted and unweighted ranges.

[Peter]

(In [2155]) Spearman now works with mixed iterators. refs #498

[Peter]

(In [2156]) refs #498. testing that Gauss works with mixed iterators

[Peter]

(In [2157]) refs #498. Zscore now works with mixed iterators and also refined requirements on iterator types

[Peter]

(In [2158]) qQuantileNormalizer now works with mixed iterators. closes #498