[BUG] Fix if_else series naming from predicate broadcast

[colin-ho]

When predicate is broadcasted in the if_else kernel and the predicate is false, the if_false series is cloned and the name is unchanged, which causes a

ValueError: DaftError::ComputeError Mismatch of expected expression name and name from computed series

at expression evaluation time because the expected field name is the if_true series name.

Example code to reproduce:

df = daft.from_pydict({"predicate": [False], "if_true": ["true"], "if_false": ["false"]})
df = df.select((df["predicate"] == True).if_else(df["if_true"], df["if_false"]))
df.show()