Function to cast InferenceData into tidy_draws format

[damonbayer]

tidy_draws details.

You can approach this however you like, but I have previously done some relevant work:

I have written similar functionality in R, which expects as input the output of arviz.InferenceData.to_dataframe.
It splits the InferenceData into several tidy_draws formatted tables, one for each group.

arviz_split <- function(x) {
x %>%
    select(-distribution) %>%
    split(f = as.factor(x$distribution))
}

pyrenew_samples <-
read_csv(inference_data_path) %>%
rename_with(\(varname) str_remove_all(varname, "\\(|\\)|\\'|(, \\d+)")) |>
rename(
    .chain = chain,
    .iteration = draw
) |>
mutate(across(c(.chain, .iteration), \(x) as.integer(x + 1))) |>
mutate(
    .draw = tidybayes:::draw_from_chain_and_iteration_(.chain, .iteration),
    .after = .iteration
) |>
pivot_longer(-starts_with("."),
    names_sep = ", ",
    names_to = c("distribution", "name")
) |>
arviz_split() |>
map(\(x) pivot_wider(x, names_from = name) |> tidy_draws())

[AFg6K7h4fhy2]

A comment is due on why this issue was modified from High Priority to Medium Priority during the week of [Oct. 28, Nov. 1]. Given DB's workflow (see partially here), which loads in R the contents of an az.InferenceData object, in pyrenew-hew (see here), careful treatment of time intervals in az.InferenceData was needed more (see here) by DB, who with the rest of STF represents the primary user of forecasttools-py, for the time being.