add message that datatable.nomatch will be deprecated

[jangorecki]

Closes #3585 (revised)
First step is to inform and communicate.

[codecov]

Codecov Report

Merging #3612 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #3612      +/-   ##
==========================================
+ Coverage   97.81%   97.81%   +<.01%     
==========================================
  Files          66       66              
  Lines       12927    12929       +2     
==========================================
+ Hits        12645    12647       +2     
  Misses        282      282

Impacted Files	Coverage Δ
R/foverlaps.R	98.11% <100%> (+0.01%)	⬆️
R/data.table.R	97.74% <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 daf721a...88ac189. Read the comment docs.

[codecov]

Codecov Report

Merging #3612 into master will not change coverage.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master    #3612   +/-   ##
=======================================
  Coverage   98.19%   98.19%           
=======================================
  Files          66       66           
  Lines       12922    12922           
=======================================
  Hits        12689    12689           
  Misses        233      233

Impacted Files	Coverage Δ
R/foverlaps.R	98.1% <100%> (ø)	⬆️
R/data.table.R	97.74% <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 06a5ed6...21e12cd. Read the comment docs.

[mattdowle]

Not sure if you saw my earlier comment here : #3578 (comment). This would allow us to not introduce the new datatable.natural option. And we'd deprecate datatable.nomatch too. There would be no need to document unsafe options, because there wouldn't be any unsafe options.

[jangorecki]

Yes, but it is less trivial, and need message about nomatch option anyway. Deprecating nomatch will takes ages, this PR is first step. I can remove naturaljoin option altogether in this PR.

[mattdowle]

Yes deprecating nomatch option would need time and communication, and a message would be first step.

I can remove naturaljoin option altogether in this PR.

Do you mean you agree with my suggestion to use on=data.table to trigger natural join, or some other way to avoid the option?

[jangorecki]

I agree but I don't want to include that in this PR. It would be good to have it in separate PR. This would be API change and once users will start to use it will be difficult to amend changes to it.