between NA bounds, nanotime support

[jangorecki]

closes #3667
For upper/lower argument of vector type (non-scalar) could be better optimised. IMO it is not a priority as we ultimately need to push down character handling into C. Which was also mentioned in comment

TODO: support character between in Cbetween and remove this branch

This PR at least put expected functionality and unit tests in place so when we proceed to C development we have a properly working R code to match to.

---

also closes #3522 by fallback to R code

---

also added nanotime support

[codecov]

Codecov Report

Merging #3731 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #3731      +/-   ##
==========================================
+ Coverage   99.41%   99.41%   +<.01%     
==========================================
  Files          71       71              
  Lines       13199    13210      +11     
==========================================
+ Hits        13122    13133      +11     
  Misses         77       77

Impacted Files	Coverage Δ
src/utils.c	100% <100%> (ø)	⬆️
R/between.R	100% <100%> (ø)	⬆️
src/init.c	100% <100%> (ø)	⬆️
src/between.c	100% <100%> (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 a8e926a...3bf1d02. Read the comment docs.

[jangorecki]

failure of CI is quite interesting.
github is automerging this branch into master, and CI is testing some artificial merge commit: ea7d608
at first glance it looks like a bug
gitlab CI handles that as expected: https://gitlab.com/jangorecki/data.table/pipelines/74562885

[mattdowle]

This one was tricky! Changes I made were

• NAbounds implemented at C level
• Supported character at C level
• Changed new NAbounds= from TRUE/FALSE to TRUE/NA. The idea being that TRUE/FALSE is hard to remember which way around it means. But if you remember it as what is returned for that bound when there's an NA, then TRUE/NA hopefully is easier to remember. Explained in the news item and man page too, using the adjective 'unlimited' bound.
• Rinherits() internal C function created which observes S4 inheritance from integer64 (nanotime). Enabled removing special cases for nanotime. Might be able to remove nanotime special cases elsewhere too now using this.
• Moved INHERITS to utils.c to be next to Rinherits(). I'm not comfortable with their names (e.g. all-caps doesn't give much of clue) but I couldn't think what would be better names right now. They are internal functions so we can change their names anytime.