flatten_dfr drops all elements except the first

[expersso]

Consider this:

> nrow(iris)
[1] 150
> nrow(flatten_dfr(list(iris, iris, iris)))
[1] 150
> nrow(bind_rows(iris, iris, iris))
[1] 450

I would expect flatten_dfr to simply concatenate the data frame like bind_rows, but it appears to just drop all elements except the first.

[batpigandme]

Assume it's the same reason the flatten test is failing. (So, test works, that's good news!)

  > test_check("purrr")
  [31m──[39m [31m1. Failure: can flatten to a data frame with named lists [39m
  `flatten_dfc(list(1))` did not throw an error.
  
  ══ testthat results  ═══════════════════════════════════════
  OK: 433 SKIPPED: 0 FAILED: 1
  1. Failure: can flatten to a data frame with named lists (@test-flatten.R#78) 

[lionel-]

Using rlang::flatten() instead of purrr::flatten() would fix this because it only flattens inner lists. For instance, the following all return the same results:

list(mtcars, mtcars))
rlang::flatten(list(mtcars, mtcars)))           # No inner list
rlang::flatten(list(mtcars, list(mtcars))))     # Flattened inner list

We're still not sure what the proper flattening semantics are...

[hadley]

This seems to be the same problem as #757 — in this case flatten_dfr() appears to be column-binding, not row-binding.