Add to_binary methods for integer types

[mcrumiller]

Description

See #21549 (comment).

Currently, casting most data types to binary goes through pl.String, as in:

>>> x = pl.Series([0, 2**16-1], dtype=pl.UInt16)
>>> x.cast(pl.Binary)
shape: (2,)
Series: '' [binary]
[
        b"0"
        b"65535"
]

Numpy allows using .tobytes() or .view(np.void) to convert to a binary view:

>>> x = np.array([0, 2**16-1], dtype=np.uint16)
>>> x.tobytes()
b'\x00\x00\xff\xff'
>>> x.view(np.void)
array([b'\x00\x00', b'\xFF\xFF'], dtype='|V2')

It would very useful in many contexts that require binary formats to convert series to binary, as in:

>>> x = pl.Series([0, 2**16-2], dtype=pl.UInt16)
>>> x.to_binary()
shape: (2,)
Series: '' [binary]
[
        b"0000"
        b"FEFF"
]

As an alternative, Expr.reinterpret could be enhanced to allow for pl.Binary, although the input arguments would probably have to be amended.