`unnest` does not work with `tsibble` being the elements of a column

[FinYang]

library(tidyverse)
library(tsibble)
#> 
#> Attaching package: 'tsibble'
#> The following object is masked from 'package:lubridate':
#> 
#>     interval
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, union
test <- tibble(a=1, b=list(tsibble(i = 1:10, x = rnorm(10), index = i)))
test
#> # A tibble: 1 × 2
#>       a b                
#>   <dbl> <list>           
#> 1     1 <tbl_ts [10 × 2]>
unnest(test, b)
#> Error in `validate_index()`:
#> ! Column `i` (index) must not contain `NA`.
#> Backtrace:
#>      ▆
#>   1. ├─tidyr::unnest(test, b)
#>   2. └─tidyr:::unnest.data.frame(test, b)
#>   3.   └─tidyr::unchop(...)
#>   4.     └─tidyr:::df_unchop(...)
#>   5.       └─vctrs::list_unchop(col, ptype = col_ptype)
#>   6.         └─vctrs (local) `<fn>`()
#>   7.           ├─vctrs:::vec_restore_dispatch(x = x, to = to)
#>   8.           └─tsibble:::vec_restore.tbl_ts(x = x, to = to)
#>   9.             └─tsibble::build_tsibble(...)
#>  10.               └─tsibble:::validate_index(tbl, !!qindex)
#>  11.                 └─rlang::abort(...)
unnest_tsibble(test, b)
#> Error in `validate_index()`:
#> ! Column `i` (index) must not contain `NA`.
#> Backtrace:
#>      ▆
#>   1. └─tsibble::unnest_tsibble(test, b)
#>   2.   ├─tidyr::unnest(as_tibble(data), cols = !!cols)
#>   3.   └─tidyr:::unnest.data.frame(as_tibble(data), cols = !!cols)
#>   4.     └─tidyr::unchop(...)
#>   5.       └─tidyr:::df_unchop(...)
#>   6.         └─vctrs::list_unchop(col, ptype = col_ptype)
#>   7.           └─vctrs (local) `<fn>`()
#>   8.             ├─vctrs:::vec_restore_dispatch(x = x, to = to)
#>   9.             └─tsibble:::vec_restore.tbl_ts(x = x, to = to)
#>  10.               └─tsibble::build_tsibble(...)
#>  11.                 └─tsibble:::validate_index(tbl, !!qindex)
#>  12.                   └─rlang::abort(...)

# Expected outcome
test %>%
  mutate(b = lapply(b, as_tibble)) %>%
  unnest(b) %>%
  as_tsibble(index = i, key = a)
#> # A tsibble: 10 x 3 [1]
#> # Key:       a [1]
#>        a     i       x
#>    <dbl> <int>   <dbl>
#>  1     1     1  0.0625
#>  2     1     2 -0.173 
#>  3     1     3 -0.283 
#>  4     1     4  2.33  
#>  5     1     5  0.181 
#>  6     1     6  0.340 
#>  7     1     7 -0.0122
#>  8     1     8 -0.988 
#>  9     1     9  0.827 
#> 10     1    10  0.679

^{Created on 2023-04-11 with reprex v2.0.2}

If the output cannot be a tsibble since the method is on tibble, can it be a working tibble? Or at least with a more meaningful error message?