Clarification of error message on reading CSVs in large vignette

[mine-cetinkaya-rundel]

In the large vignette, don't fully understand what the second information point is saying:

readr::read_csv("a\n1", show_col_types = FALSE) |>
  as_duckdb_tibble()
#> Error in `as_duckdb_tibble()` at duckplyr/R/ducktbl.R:81:3:
#> ! The input is data read by readr, and duckplyr supports
#>   reading CSV files directly.
#> ℹ Use `read_csv_duckdb()` to read with the built-in reader.
#> ℹ To proceed with the data as read by readr, use `as_tibble()` before
#>   `as_duckdb_tibble()`.

When I read in with readr() it's already a tibble, why would I need to " use as_tibble() before as_duckdb_tibble()"?

[krlmlr]

The rationale here is that in a duckplyr driven pipeline, using DuckDB to read CSVs is likely to be much more efficient, to the extent that the CSV may never be even fully parsed and loaded into memory. This asks the analyst to be explicit about their choice.

Extending the vignette text.