Use _RO accessors for const targets

[MichaelChirico]

Searched the sources:

grep -Er "const\b.*[*]\w+\s*=\s*[A-Z]+([^OR]|[^A]R)[(]" src

This is some low-hanging fruit in terms of better use of read-only accessors. There will be other cases where we can make the change to a const pointer as well, but those can't be identified with a simple regex.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.02%. Comparing base (e13ea60) to head (9b034c8).
⚠️ Report is 8 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #7611      +/-   ##
==========================================
- Coverage   99.02%   99.02%   -0.01%     
==========================================
  Files          87       87              
  Lines       16903    16890      -13     
==========================================
- Hits        16739    16726      -13     
  Misses        164      164              

☔ 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.

[github-actions]

No obvious timing issues in HEAD=ro-access

Generated via commit 9b034c8

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

Task	Duration
R setup and installing dependencies	3 minutes and 0 seconds
Installing different package versions	43 seconds
Running and plotting the test cases	4 minutes and 26 seconds

[MichaelChirico]

Another big class of cases where we could switch to _RO are call sites like INTEGER(x)[k] -- once we do [ to the output, we know we're only reading from x.

There are roughly 240 such sites (grep -Er "(INTEGER|LOGICAL|REAL|COMPLEX|RAW)[(]\w+[)]\[[^=]*$"), I just suspect it's not worth it to change them.

[MichaelChirico]

I see this has some (or total) overlap with #7394. I think it's good to break out the changes into minimally-digestible parts -- here we focus only on const <type> *<var> = <writeable>(x) --> const <type> *<var> = <read-only>(x).

[ben-schwen]

Should we add an accessor for integer64?

#define INTEGER64_RO(x)	((const int64_t *) DATAPTR_RO(x))

[MichaelChirico]

Should we add an accessor for integer64?

#define INTEGER64_RO(x)	((const int64_t *) DATAPTR_RO(x))

Good point, though I think out of scope -- #7618

[aitap]

This Coccinelle spatch should at least help as a linter: @@ type T; const T *variable; expression E; @@ - variable = REAL(E) + variable = REAL_RO(E) @@ type T; const T *variable; expression E; @@ - variable = INTEGER(E) + variable = INTEGER_RO(E) @@ type T; const T *variable; expression E; @@ - variable = COMPLEX(E) + variable = COMPLEX_RO(E) @@ type T; const T *variable; expression E; @@ - variable = RAW(E) + variable = RAW_RO(E) @@ type T; const T *variable; expression E; @@ - variable = LOGICAL(E) + variable = LOGICAL_RO(E)