Time series returns a warning message, "subscript out of bounds", but not in all cases

Question

Working on a time series of monthly labor data in the USA. If the forecast is set to 1 month, then this returns a warning message, but 3 months does not return a warning message.

Get the data, set the parameters:

time_series_data <- read.csv('https://raw.githubusercontent.com/InfiniteCuriosity/forecasting_jobs/main/Total_Nonfarm.csv')
train_amount <- 0.60
fc <- 1

Here is what the data looks like:

Load packages that will be used:

library(tidyverse)
library(fpp3)

Convert time series data to a tsibble:

time_series_data <- time_series_data %>%
    dplyr::mutate(Date = tsibble::yearmonth(Label), Value = Value, Change = tsibble::difference(Value)) %>%
    dplyr::select(Date, Value, Change) %>%
    tsibble::as_tsibble(index = Date) %>%
    dplyr::slice(-c(1))

Set up train and test sets:

time_series_train <- time_series_data[1:round(train_amount*(nrow(time_series_data))),]
time_series_test <- time_series_data[(round(train_amount*(nrow(time_series_data))) +1):nrow(time_series_data),]

Create a simple model, calculate error amounts (the same warning message happens with many different time series functions, such as ETS, ARIMA, etc.)

Linear1_model = fable::TSLM(Value ~ season() + trend())

Linear1_error_list <- time_series_train %>% 
  fabletools::model(fable::TSLM(Value ~ season() + trend())) %>% 
  fabletools::forecast(h = fc) %>% 
  fabletools::accuracy(time_series_test)

If fc = 1. it runs but returns the warning message:

# A tibble: 1 × 10
  .model                                  .type    ME  RMSE   MAE   MPE  MAPE  MASE RMSSE   ACF1
  <chr>                                   <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>  <dbl>
1 fable::TSLM(Value ~ season() + trend()) Test  5998. 5999. 5998.  4.14  4.14   NaN   NaN -0.131

Warning message:
1 error encountered
[1] subscript out of bounds

but if fc = 3 it runs but is no warning message.

The same warning message is returned no matter which values are selected. For example, this does not return the NA values, but also returns the warning message:

Linear1_error_list <- time_series_train %>% 
  fabletools::model(fable::TSLM(Value ~ season() + trend())) %>% 
  fabletools::forecast(h = fc) %>% 
  fabletools::accuracy(time_series_test) %>% 
  select(.model:MAPE)
Warning message:
1 error encountered
[1] subscript out of bounds

How can this run with fc = 1 and not return a warning message?