Clean up util.num.make_symmetric_matrix_from_upper_tri and util.io_util.micro_pyawk, fix new ruff errors

[DanielYang59]

Simplify/clarify util.num.make_symmetric_matrix_from_upper_tri and add tests

Caution

make_symmetric_matrix_from_upper_tri now raise ValueError for non 3x3 array

I believe utils.num.make_symmetric_matrix_from_upper_tri is designed to work on 3x3 matrix alone (the new index is hard-coded):

pymatgen/src/pymatgen/util/num.py

Lines 31 to 32 in 31f1e1f

	idx = [0, 3, 4, 1, 5, 2]
	val = np.array(val)[idx]

As such I might greatly simplify the implementation f2c3252 which gives ~3x speedup and much better readability.

Clean up util.io_util.micro_pyawk

Warning

The debug/postdebug is likely debug args, no usage across the entire code bass, schedule them for removal after one-year grace period in case someone really need them.

Others

• Remove:

  pymatgen/src/pymatgen/util/io_utils.py

  Lines 98 to 99 in 31f1e1f

  	umask = os.umask(0)
  	os.umask(umask)

  I believe this is doing nothing (set umask to 0, but reset it back the original umask immediately)

• Fix new ruff errors and bump pre-commit accordingly

[DanielYang59]

I believe this helper function is designed to work for 3x3 matrix alone, as such I might just make things simple and reject non 3x3 arrays with ValueError.

The new implementation gives a ~3 speedup:

Original implementation time: 0.378586 seconds
Simplified implementation time: 0.118705 seconds

Test script (from GPT):

import numpy as np
from timeit import timeit

def original_make_symmetric_matrix_from_upper_tri(val):
    """Original implementation."""
    idx = [0, 3, 4, 1, 5, 2]
    val = np.array(val)[idx]
    mask = ~np.tri(3, k=-1, dtype=bool)
    out = np.zeros((3, 3), dtype=val.dtype)
    out[mask] = val

    out.T[mask] = val
    return out

def simplified_make_symmetric_matrix_from_upper_tri(val):
    """Simplified implementation."""
    array = np.asarray(val)
    if array.shape != (6,):
        raise ValueError(f"Expected val of length 6, got {array.shape}")

    return np.array([
        [array[0], array[3], array[4]],
        [array[3], array[1], array[5]],
        [array[4], array[5], array[2]],
    ])

# Test input
val = [1, 2, 3, 4, 5, 6]

# Measure execution time
original_time = timeit(
    "original_make_symmetric_matrix_from_upper_tri(val)",
    globals=globals(),
    number=100000,
)
simplified_time = timeit(
    "simplified_make_symmetric_matrix_from_upper_tri(val)",
    globals=globals(),
    number=100000,
)

# Display results
print(f"Original implementation time: {original_time:.6f} seconds")
print(f"Simplified implementation time: {simplified_time:.6f} seconds")