Fix object dtype to string conversion

[RamiNoodle733]

Description

This PR fixes the issue where converting object dtype columns to string fails with a TypeError.

When calling .astype('str') on an object dtype column (e.g., numpy arrays with mixed types like [123, "test", None]), the to_string() function would fail because vaex.strings.to_string() doesn't have an overload for object dtype arrays.

Fix

Modified the to_string() function in packages/vaex-core/vaex/functions.py to handle object dtype arrays by converting them to string arrays using ar.astype(str) before passing to vaex.strings.to_string().

Changes

• Added dtype check for object arrays in to_string()
• Convert object arrays to string arrays before processing
• Handles both regular and masked arrays

Example

import vaex
import numpy as np

# This now works correctly
arr = np.array([123, "test", None], dtype=object)
df = vaex.from_arrays(test=arr)
df['test'] = df['test'].astype('str')  # No longer raises TypeError

Fixes #1686

[ddelange]

if the same thing happens for both if and else branches, it can be moved out of the if else entirely