Bootstrapped innovations with combination models

Hi,

I'm encountering an error when trying to generate simulations from a combined model with bootstrapped innovations. Is this normal due to some peculiarities of combination models or is there a bug?

Thank you very much.

x <- tsibble::tourism %>% filter(Region == "Adelaide")

# Working example
fit <- x %>% 
  model(
    ets = ETS(Trips),
    ar = ARIMA(Trips)
  )

sims <- fit %>%
     generate(times = 10, h = 4, bootstrap = T) %>% as_tibble()

# Not working
fit <- x %>% 
  model(
    ets = ETS(Trips),
    ar = ARIMA(Trips)
  ) %>% mutate(comb = (ets + ar)/2)

sims <- fit %>%
     generate(times = 10, h = 4, bootstrap = T) %>% as_tibble()

Erreur dans sample.int(length(x), size, replace, prob) : 
  premier argument incorrect
De plus : Message d'avis :
Dans mean.default(res, na.rm = TRUE) :
  l'argument n'est ni numérique, ni logique : renvoi de NA

> sessionInfo()
R version 4.1.0 (2021-05-18)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19043)

Matrix products: default

locale:
[1] LC_COLLATE=French_Canada.1252  LC_CTYPE=French_Canada.1252    LC_MONETARY=French_Canada.1252 LC_NUMERIC=C                  
[5] LC_TIME=French_Canada.1252    

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

other attached packages:
 [1] lubridate_1.7.10  concordance_2.0.0 ggplot2_3.3.5     feasts_0.2.2      vroom_1.5.4       dplyr_1.0.7       fable_0.3.1       fabletools_0.3.1 
 [9] tsibble_1.0.1     stringr_1.4.0    

loaded via a namespace (and not attached):
 [1] Rcpp_1.0.7           urca_1.3-0           progressr_0.8.0      pillar_1.6.2         compiler_4.1.0       tools_4.1.0          digest_0.6.27       
 [8] bit_4.0.4            lattice_0.20-44      nlme_3.1-152         lifecycle_1.0.0      tibble_3.1.3         gtable_0.3.0         anytime_0.3.9       
[15] pkgconfig_2.0.3      rlang_0.4.11         cli_3.0.1            withr_2.4.2          generics_0.1.0       vctrs_0.3.8          nnet_7.3-16         
[22] bit64_4.0.5          grid_4.1.0           tidyselect_1.1.1     glue_1.4.2           R6_2.5.0             fansi_0.5.0          distributional_0.2.2
[29] farver_2.1.0         purrr_0.3.4          tidyr_1.1.3          tzdb_0.1.2           magrittr_2.0.1       scales_1.1.1         ellipsis_0.3.2      
[36] colorspace_2.0-2     renv_0.14.0          utf8_1.2.2           stringi_1.7.3        munsell_0.5.0        crayon_1.4.1  

Run the provided tourism example and compare bootstrap generation for the separate ETS and ARIMA models with the combined (ets + ar)/2 model. Start at the generate() entry point and trace how bootstrapped innovations are handled for combination models; done means the combined example generates simulations without the reported error, or the limitation is clearly documented.