fix: handle MATLAB cell arrays with empty/nested elements

[dimitri-yatsenko]

Summary

Fix read_cell_array to handle edge cases from MATLAB that caused ValueError: cannot reshape array errors.

Problem

MATLAB cell arrays can contain:

• Empty elements: {[], [], []}
• Nested arrays of different sizes: {[1,2,3], [4,5,6,7,8]}
• Mixed empty/non-empty elements

The previous implementation used np.array(result).reshape(shape) which fails because:

1. NumPy can't create uniform arrays from ragged data
2. Empty arrays have size 0 and can't reshape to (3,1)

Solution

Use dtype='object' to avoid NumPy's array homogeneity requirements:

# Before (fails on ragged arrays)
return np.array(result).reshape(shape, order="F")

# After (handles all cases)
arr = np.empty(n_elem, dtype=object)
arr[:] = result
return arr.reshape(shape, order="F")

Test plan

• Added tests for nested arrays (issue Cell array of arrays of doubles cannot be fetched in python, only in matlab #1098)
• Added tests for empty elements (issue Read an empty cell array blob fails #1056)
• Added tests for mixed empty/non-empty elements
• Existing MATLAB blob tests still pass

Closes #1056
Closes #1098

---

🤖 Generated with Claude Code