Optimize, standardize, and add tests and coverage reports

[patrickbarks]

(1) Favour slot access with @ internally
It's substantially faster. Accessors aren't really slowing down any of our functions, but particularly for 'lower-level' functions that may be called repeatedly in users' loops or vectorized functions (e.g. [, $, matA()), I think it's a good idea to optimize.

(2) Add methods for all mat-related accessors (matA(), matU(), ..., matrixClass(), etc.) for signature 'list', to enable e.g. matA(db$mat). The purpose is to make it easier to work with the tidyverse, e.g.

db_tidy <- fetchDB("Compadre") %>% 
  CompadreData() # extract data slot (just a tibble)

db_tidy %>% 
  mutate(stage_surv = lapply(matU(mat), colSums)) # use matU on list-column


# without the list method, could only use matU() within a custom fn, e.g.
db_tidy %>% 
  mutate(stage_surv = lapply(mat, function(x) colSums(matU(x))))

(3) Re-incorporate @levisc8's unit tests, and expand to all fns (toward #13). Also added code coverage checks via covr, to check how much of our codebase is covered by unit tests. Results will be shown in badge on README. Will also give coverage reports for pull requests.

Other edits:

• remove convertLegacyDB and replace use with asCompadreDB (they do the same thing)
• add CompadreLegacy subsample to distribute with package (for examples and testing asCompadreDB)
• further optimization to validCompadreMat checks (down to 3.5 s for COMPADRE X.X.X)
• edit [ method to allow column drop with db[,-j] (still always retains mat column)
• add examples to ?CompadreDB doc
• update and standardize documentation
• change data.R doc to raw markdown to enable better formatting (in my opinion)
• revert slot names to data and version (per my suggesion in rename "data" generic? #37)
• added Contributor Code of Conduct (required by rOpenSci, and good idea generally)

[jonesor]

Thanks @patrickbarks. Looks great!