fifelse() coerces NA to other types and supports vectorized na argument

[shrektan]

Closes #4287
Closes #4286
Closes #4277

Two points:

1. na supports vectorized input: The na argument should be no differences from yes and no due to the fact that TRUE, FALSE and NA are all valid R's logical values
2. The logical value of NAwill be coerced to other types automatically:
   1. The na argument already performs this behavior. As stated in the 1 point, yes and no should behave the same;
   2. data.table does this auto-coercion in many other functions
   3. Should only have a very small performance impact

[codecov]

Codecov Report

Merging #4289 (0c810d2) into master (9847445) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #4289   +/-   ##
=======================================
  Coverage   99.47%   99.47%           
=======================================
  Files          75       75           
  Lines       14726    14775   +49     
=======================================
+ Hits        14648    14697   +49     
  Misses         78       78           

Impacted Files	Coverage Δ
src/fifelse.c	100.00% <100.00%> (ø)

---

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 9847445...0c810d2. Read the comment docs.