Use `str` Instead of `repr` for Data Value Serialization

[mandli]

The data_write function in data.py uses repr to serialize data in the data objects currently. While this often produces the same output as str (str depends on repr) it does not always produce what needs to be output and read in by Fortran code. This was found while trying to switch some paths to the newer pathlib library

from pathlib import Path
my_path = Path() / "my_data.txt"

so that repr(my_path) produces

"PosixPath('my_data.txt')"

where as str(my_path) or f"{my_path}" produces

'my_data.txt'

Note that this is the correct behavior of repr and str, we have just been using them incorrectly.

[mandli]

I should add that numpy arrays also do not serialize correctly, although it is not a problem with repr. Current code converts a np.ndarray to a list and then tries to serialize with repr such that

import numpy as np
x = np.array([1, 2, 3])
repr(list(x))

produces '[np.int64(1), np.int64(2), np.int64(3)]'. Note that str does not help here, instead we should probably use str(x), which directly produces

'[1 2 3]'

The repr(x) produces something that is not correct here too

 'array([1, 2, 3])'