`df.join` should accept singleton arguments, not just iterable.

[thomcom]

join uses forEach to join across the on: keys. A common use case of join is with a single join column, so we should accept the use of a single join argument.

keys_df.join({other: target_df, on: 'keys', how: 'inner'});

vs

keys_df.join({other: target_df, on: ['keys'], how: 'inner'});