Refactor VCA

[rguliev]

• Remove seed from function arguments (see here)
• Make output similar to N-FINDR: return only indices which can be converted later by endmembers() method using specific data
• Move dimensionalityReduction out of VCA? In current implementation, dimension reduced data is not accessible. N-FINDR approach is to move dimension reduction outside to: 1) make it explicit which data is used, 2) make the data accessible for later use, 3) make faster re-runs with different parameters.
• Rename dimensionalityReduction to make it explicit that it is designed only for VCA. How about vca_dr?
• Similar for estSNR. This method is designed for VCA. I don't think it can and should be generalized. Perhaps, it's also better to rename it to something like vca_snr
• Remove SNR as function arguments for vca05 and vcaLopez2012. It seems like this argument is not needed and SNR only used in dimensionalityReduction
• Use consistent names for VCA implementations. Using analogy with N-FINDR it could be vca_nascimento and vca_lopez

[rguliev]

@bryanhanson, @cbeleites should we discuss it?

[bryanhanson]

I will look at this w/I 24 hrs.

[cbeleites]

looks good to me 👍

[bryanhanson]

Agreed, looks good. Consistency and simplicity! Thanks for working on this!