Unify messages for translation

[MichaelChirico]

Gemini was used to identify the similar strings; I read over the analysis and applied its findings as appropriate.

This should help out the translation teams by minimizing duplicate work.

Details

Exact prompt to Gemini:

@po/R-data.table.pot is a translation template file. Find similar strings in it that can be unified.

   - First, use the shell to remove all extraneous information from the files and only look at the distinct msgid entries, excluding the initial header entry.
   - Second, examine the corpus of strings and pull out ones that should be unified. Some examples of what we're after:
     + "hello '%s'" and "hello '%s'." could unify by normalizing punctuation
     + "x is negative" and "y is negative" could unify by adding a template: "%s is negative"
     + "bad '%s' argument" and "invalid '%s' argument" could unify by aligning on consistent language: "invalid '%s' argument"
     + "bad '%s' argument" and 'bad "%s" argument' could unify by aligning on consistent punctuation: "bad '%s' argument"
     + There are many other types of messages that could be unified -- use your judgment.
   - Some _negative_ examples that are ineligible:
     + "DLL %s not found" and "shared object %s not found" _look_ like they could unify as "%s %s not found", but there might be languages where "shared object" cannot be pulled out to the template with fidelity
     + "invalid input" and "invalid named string" _look_ like they could unify as "invalid %s", but again, this only works in English, not for generic target languages

[github-actions]

• HEAD=i18n-unify stopped early for DT[by,verbose=TRUE] improved in #6296
• HEAD=i18n-unify stopped early for DT[,.SD] improved in #4501
  

Generated via commit ae793fb

Download link for the artifact containing the test results: ↓ atime-results.zip

Task	Duration
R setup and installing dependencies	2 minutes and 49 seconds
Installing different package versions	44 seconds
Running and plotting the test cases	4 minutes and 22 seconds

[codecov]

Codecov Report

❌ Patch coverage is 94.44444% with 4 lines in your changes missing coverage. Please review.
✅ Project coverage is 98.96%. Comparing base (3e579ee) to head (ae793fb).

Files with missing lines	Patch %	Lines
R/data.table.R	85.71%	1 Missing ⚠️
src/assign.c	0.00%	1 Missing ⚠️
src/openmp-utils.c	0.00%	1 Missing ⚠️
src/uniqlist.c	0.00%	1 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #7562   +/-   ##
=======================================
  Coverage   98.96%   98.96%           
=======================================
  Files          87       87           
  Lines       16737    16739    +2     
=======================================
+ Hits        16563    16565    +2     
  Misses        174      174           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[ben-schwen]

Suggested change

	stopf("x is a list, 'cols' cannot be 0-length.")
	stopf("x is a list, 'cols' cannot be 0-length")

since "most" error messages dont end with ., but NIT

[ben-schwen]

Basically LGTM.

Currently we also two different type of styles whether error message end with . or not. From a quick look, this mostly depends on the source file.