[Python] mix of np.float16/np.float128 and integers converts to Int64Array, not DoubleArray

### Describe the bug, including details regarding any error messages, version, and platform.

When you mix `np.float32` or `np.float64` scalars and integers, the result is `DoubleArray`:

```python
>>> pa.array([np.float32(0.5), 1])

[
0.5,
1
]
>>> pa.array([np.float64(0.5), 1])

[
0.5,
1
]
```

But when you mix `np.float16` or `np.float128` scalars and integers, the result is `Int64Array`, and the floats are truncated to integers:

```python
>>> pa.array([np.float16(0.5), 1])

[
0,
1
]
>>> pa.array([np.float128(0.5), 1])

[
0,
1
]
```

### Component(s)

Python