fix for melt with na.rm=TRUE and list column value

[tdhock]

hi again, I have been trying to get 100% coverage in fmelt.c and some of the uncovered lines revealed a bug. This PR adds a test for this case, and will eventually add a fix. The problem is shown below:

library(data.table)
melt(data.table(a1=1, b1=1, b2=2), na.rm=TRUE, measure.vars=list(a="a1", b=c("b1","b2")))
#>    variable a b
#> 1:        1 1 1
# only one row as expected (since a2 is missing, b2 is removed as well).
melt(data.table(a1=1, b1=list(1:2), b2=list(c('foo','bar'))), na.rm=TRUE, measure.vars=list(a="a1", b=c("b1","b2")))
#>    variable  a       b
#> 1:        1  1     1,2
#> 2:        2 NA foo,bar
# two rows, second a value missing, incorrect since na.rm=TRUE was specified.

[tdhock]

There was some code which treated lists as a special case, and always used data->narm=FALSE in the C code, even when the user specified na.rm=TRUE in the R code. Seems to me that this is undesirable, due to the inconsistency with the non-list case. If that behavior is desirable to keep, then I would have at least expected some kind of message telling the user that their request of na.rm=TRUE was ignored (seems like there was a message but only if verbose=TRUE).

[codecov]

Codecov Report

Merging #4737 (cb510cf) into master (d8e0fb3) will increase coverage by 0.01%.
The diff coverage is n/a.

@@            Coverage Diff             @@
##           master    #4737      +/-   ##
==========================================
+ Coverage   99.45%   99.46%   +0.01%     
==========================================
  Files          73       73              
  Lines       14612    14609       -3     
==========================================
- Hits        14532    14531       -1     
+ Misses         80       78       -2     

Impacted Files	Coverage Δ
src/fmelt.c	100.00% <ø> (+0.35%)	⬆️

---

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 d8e0fb3...cb510cf. Read the comment docs.

[tdhock]

This behavior is mentioned on the ?melt.data.table man page:

# return 'NA' for missing columns, 'na.rm=TRUE' ignored due to list column
melt(DT, id=1:2, measure=patterns("l_", "c_"), na.rm=TRUE)

[mattdowle]

I guess the news item in #4720 covers this one too.