Improved intensity pairs trying to make spectrum library comparisons faster

[trishorts]

The primary purpose of this PR is to make the computation of spectrum similarity a little faster.
Overall, I attempted also to improve the code.

Some points to clarify because some changes were made:

1. We assume in general that we'll be comparing experimental to theoretical. We wish to keep all theoretical peaks even if they are not paired to an experimental because that will lower the similarity as desired. Default is to keep all theoretical peaks.
2. If desired, the user can choose to use all experimental peaks. Default however is to not.
3. To enable comparison of two experimental spectra, we now have two separate bools for keeping unpaired peaks.
4. Lots of explanations were added to unit tests so that future understanders can understand.
5. KullbackLeibler, which should be used for isotope envelopes, is improved but unchanged.
6. We are now officially tossing ALL zero intensity peaks from the start. doesn't matter, experimental or theoretical.

[codecov]

Codecov Report

Attention: Patch coverage is 94.23077% with 6 lines in your changes are missing coverage. Please review.

Project coverage is 75.04%. Comparing base (cf854ac) to head (400c865).

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #765      +/-   ##
==========================================
- Coverage   75.06%   75.04%   -0.02%     
==========================================
  Files         189      189              
  Lines       30235    30236       +1     
  Branches     3035     3034       -1     
==========================================
- Hits        22697    22692       -5     
- Misses       7024     7029       +5     
- Partials      514      515       +1     

Files	Coverage Δ
mzLib/Omics/SpectrumMatch/LibrarySpectrum.cs	93.22% <100.00%> (ø)
...b/MassSpectrometry/MzSpectra/SpectralSimilarity.cs	97.55% <94.11%> (-2.10%)	⬇️