dcast on missing int64 value generates 9218868437227407266 instead of NA #4561

When I dcast a data.table that contains an int64 column, it fills in 9218868437227407266 for any missing values instead of the usual NA.

This is related to Issues #488, #1385, and #3723, but none of those fixes have resolved this error from arising when using dcast(). This has been asked on StackOverflow here (no MWE and no response) and here where the latter answer simply proposed a workaround, but should have been filed as an issue.

# Minimal reproducible example

library(data.table)
apple <- fread("id, time, y
               a, 1, 12345678901234
               b, 1, 70
               b, 2, 20")
dcast(apple, id ~ time, value.var = "y")
   id              1                   2
1:  a 12345678901234 9218868437227407266
2:  b             70                  20

# Output of sessionInfo()

R version 4.0.1 (2020-06-06)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 20.04 LTS

Matrix products: default
BLAS: /usr/lib/x86_64-linux-gnu/blas/libblas.so.3.9.0
LAPACK: /usr/lib/x86_64-linux-gnu/lapack/liblapack.so.3.9.0

locale:
[1] LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C LC_TIME=en_US.UTF-8
[4] LC_COLLATE=en_US.UTF-8 LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
[7] LC_PAPER=en_US.UTF-8 LC_NAME=C LC_ADDRESS=C
[10] LC_TELEPHONE=C LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] data.table_1.12.9 RPostgres_1.2.0

loaded via a namespace (and not attached):
[1] compiler_4.0.1 bit_1.1-15.2 hms_0.5.3 tools_4.0.1 DBI_1.1.0
[6] Rcpp_1.0.4.6 bit64_0.9-7 vctrs_0.3.1 blob_1.2.1 pkgconfig_2.0.3
[11] rlang_0.4.6