[R] Fix excess "Handling string data with embedded nuls" warning in tests

[asfimport]

The R tests raise a warning

══ Warnings ════════════════════════════════════════════════════════════════════════════════════════════════
1. Handling string data with embedded nuls (test-RecordBatch.R:547:3) - Stripping '\0' (nul) from character vector

That test is already has an expectation for the warning, so it is likely being raised repeatedly and the excess are not suppressed.

AC: Warning not arising in tests because it is handled in an appropriate manner

Reporter: Edward Visel / @alistaire47

Related issues:

• [R] unique() and is.na() on a column of a tibble is much slower after writing to and reading from a parquet file (relates to)

_{Note: This issue was originally created as ARROW-16188. Please see the migration documentation for further details.}

[asfimport]

Jonathan Keane / @jonkeane:
FWIW the official answer from testthat is to use suppressWarnings() in this kind of case: r-lib/testthat#1083

[asfimport]

Neal Richardson / @nealrichardson:
I believe this is related to when we added ALTREP support, which changed when this warning occurs. But before we suppress the warning, we should make sure we're not doing something unexpected, like converting the data twice.

[asfimport]

Neal Richardson / @nealrichardson:
I believe this is caused by the issue raised on ARROW-16578.

[thisisnic]

I don't recall seeing these issues on our CI recently, so I'm going to assume this is fixed - feel free to reopen if not.