frame.assign.apply_element dtype argument not working when selection is multiple columns

[chaburkland]

Description

The dtype argument to sf.Frame.assign.apply_element doesn't behave as expected when the selection contains multiple columns.

Example

from datetime import date
import static_frame as sf

frame = sf.Frame.from_records(
    [
        ["2023-01-01", "2023-02-01"],
        ["2023-01-02", "2023-02-02"],
        ["2023-01-03", "2023-02-03"],
    ],
    columns=["A", "B"],
)

single = frame.assign["A"].apply_element(date.fromisoformat, dtype="datetime64[D]")
multiple = frame.assign[["A"]].apply_element(date.fromisoformat, dtype="datetime64[D]")

assert single['A'].dtype == 'datetime64[D]'
assert multiple['A'].dtype == object

Platform

17:15:58 Out [13] 
<Series: platform>
<Index>
platform           Linux-5.4.0-156-generic-x86_64-with-glibc2.29
sys.version        3.8.12 (default, Aug 30 2023, 21:17:56) [GCC 9.4.0]
static-frame       1.6.0
numpy              1.24.3
pandas             1.5.3
xlsxwriter         3.1.0
openpyxl           3.0.9
xarray             <ModuleNotFoundError>
tables             3.8.0
pyarrow            0.17.1
msgpack            (1, 0, 0)
msgpack_numpy      None
<<U13>             <object>