Update data.table dependency to version 1.14.1 once memory leakage problem is fixed

[hongyuanjia]

See Rdatatable/data.table#3292. Since reading IDF is a quite common operation in eplusr, this makes R memory usage grow larger and larger.

dt <- data.table::data.table(letter = sample(LETTERS, 1e6, TRUE))
data.table::fwrite(dt, "test.csv")

for (i in 1:10) {
    cat("Before: ", sep = "")
    print(pryr::mem_used())
    dt <- data.table::fread("test.csv", header = FALSE)
    rm(dt)
    gc()
    cat("After: ", sep = "")
    print(pryr::mem_used())
}
#> Before:
#> Registered S3 method overwritten by 'pryr':
#>   method      from
#>   print.bytes Rcpp
#> 53.5 MB
#> After: 46.8 MB
#> Before: 46.8 MB
#> After: 47.6 MB
#> Before: 47.6 MB
#> After: 48.4 MB
#> Before: 48.4 MB
#> After: 49.2 MB
#> Before: 49.2 MB
#> After: 50 MB
#> Before: 50 MB
#> After: 50.8 MB
#> Before: 50.8 MB
#> After: 51.6 MB
#> Before: 51.6 MB
#> After: 52.4 MB
#> Before: 52.4 MB
#> After: 53.2 MB
#> Before: 53.2 MB
#> After: 54 MB

for (i in 1:10) {
    cat("Before: ", sep = "")
    print(pryr::mem_used())
    df <- read.csv("test.csv", header = FALSE)
    rm(df)
    gc()
    cat("After: ", sep = "")
    print(pryr::mem_used())
}
#> Before: 54 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB
#> Before: 54.2 MB
#> After: 54.2 MB

^{Created on 2021-03-04 by the reprex package (v1.0.0)}

[microsat2022]

my data.table was 1.14.2. But it seemed that this issue was not solved.
Before: 322 MB
After: 333 MB
Before: 333 MB
After: 343 MB
Before: 343 MB
After: 354 MB
Before: 354 MB
After: 365 MB
Before: 365 MB
After: 375 MB
Before: 375 MB
After: 386 MB
Before: 386 MB
After: 396 MB
Before: 396 MB
After: 407 MB
Before: 407 MB
After: 418 MB
Before: 418 MB
After: 428 MB

[hongyuanjia]

Yes, that's expected. Unfortunately, the PR Rdatatable/data.table#3292 was still not merged in data.table v1.42.2.