Fill NAs with latest non-NA value when doing join/megre with 2 time columns

Question

Suppose I have the data like this (short one)

df_field

machine         date_start    temp_start       date_end     temp_end    
A               01/11/2