Implement g3a_otherfood_table

[lentinj]

Pumping a table of abundance from a model into another is hard(tm):

st <- g3_stock("x", 1:19)

oth_table <- expand.grid(
    year = 1991:2020,
    step = 1:4,
    length = 1:19 )
oth_table$value <- floor(runif(nrow(oth_table), 1e3, 1e4))

lookup_array <- function(tbl_name, tbl) {
    tbl_arr <- g3_array_agg(gadget3:::g3_distribution_preview(oth_table))

    out <- gadget3:::call_to_formula(substitute(
        arr[,step = g3_idx(cur_step), year = g3_idx(cur_year - start_year + 1)],
        list(arr = as.symbol(tbl_name)) ), list())
    environment(out)[[tbl_name]] <- tbl_arr
    return(out)
}

actions <- list(
  g3a_time(2000, 2005, c(3,3,3,3)),
  g3a_otherfood(st, lookup_array("oth_abund", oth_table), force_lengthvector = FALSE ),
  NULL)
full_actions <- c(actions, list(
    g3a_report_detail(actions) ))
model_fn <- g3_to_r(full_actions)
model_cpp <- g3_to_tmb(full_actions)

attr(model_cpp, 'parameter_template') |>
    identity() -> params.in
nll <- model_fn(params.in) ; r <- attributes(nll) ; nll <- as.vector(nll)

print(r$detail_x__num)

gadget3:::ut_tmb_r_compare2(model_fn, model_cpp, params.in)

What would be nice is a g3a_otherfood_table(), which munges the input table into an array as we do with likelihood data, and then fills the output.