daft.Expression.__or__ and daft.Expression.__and__ should not propagate nulls

[kevinzwang]

Describe the bug

Our expressions for or + and yield null when either side is null, but when the statement is always true or false regardless of the null-side value, we should yield that instead of propagating nulls.

To Reproduce

Consider this dataframe:

import daft

df = daft.from_pydict({
    "a": [True, False],
    "b": [None, None]
})

df.select(df["a"] | df["b"]) and df.select(df["a"] & df["b"]) both yield a column with two null rows.

Same behavior exists in SQL

Expected behavior

df.select(df["a"] | df["b"]) should yield [True, null] since the first row is always true.

df.select(df["a"] & df["b"]) should yield [null, False] since the first row is always false.

Component(s)

SQL, Other

Additional context

No response