make variance() return variance (not covariance) on matrices, closes #55

[mjskay]

This PR closes #55 via:

• Making variance.default() throw an error
• Making variance.numeric() coerce to a vector prior to calling stats::var()
• Documenting this divergence from stats::var() in the details section of the docs on variance()

Thanks for considering this and let me know if you'd like any changes.

[codecov-commenter]

Codecov Report

Merging #56 (1bff88b) into master (ab2fd9e) will increase coverage by 0.16%. The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #56      +/-   ##
==========================================
+ Coverage   62.91%   63.07%   +0.16%     
==========================================
  Files          45       45              
  Lines        1305     1308       +3     
==========================================
+ Hits          821      825       +4     
+ Misses        484      483       -1     

Impacted Files	Coverage Δ
R/distribution.R	47.10% <100.00%> (+2.19%)	:arrow_up:

---

Continue to review full report at Codecov.

Legend - Click here to learn more Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov. Last update ab2fd9e...1bff88b. Read the comment docs.

[mjskay]

Note: the CHECK failure on R 3.5 is due to ggdist not being installed, which is due to fda not working on 3.5. fda was a dependency of ggdist for a brief time but is now in Suggests (that change hasn't hit CRAN yet though). Anyway, the upshot is that the failure in R 3.5 has nothing to do with this PR (which passes on all other R versions).