feat(python): support transparent DataFrame init from numpy structured/record arrays.

[alexander-beedie]

Partially addresses #8564 (this PR adds init support; a follow-up PR will support export).
Also #3520 (comment).

HandlesDataFrame init from numpy structured/record arrays¹, recognising column names and dtypes and destructuring appropriately; it's a clean addition to the existing (internal) numpy_to_pydf construction function, so no new user-facing parameters (it will "just work").

Example

import numpy as np
import polars as pl

structured_array = np.array(
    [
        ("Google Pixel 7", 521.90, True),
        ("Apple iPhone 14 Pro", 999.00, True),
        ("Samsung Galaxy S23 Ultra", 1199.99, False),
        ("OnePlus 11", 699.00, True),
    ],
    dtype = np.dtype(
        [
            ("product", "U32"),
            ("price_usd", "float64"),
            ("in_stock", "bool"),
        ]
    ),
)

Before: (results in a single anonymous "object" column)

pl.DataFrame( structured_array )

# shape: (4, 1)
# ┌───────────────────────────────────┐
# │ column_0                          │
# │ ---                               │
# │ object                            │
# ╞═══════════════════════════════════╡
# │ ('Google Pixel 7', 521.9, True)   │
# │ ('Apple iPhone 14 Pro', 999., Tr… │
# │ ('Samsung Galaxy S23 Ultra', 119… │
# │ ('OnePlus 11', 699., True)        │
# └───────────────────────────────────┘

After: (destructured to frame)

pl.DataFrame( structured_array )

# shape: (4, 3)
# ┌──────────────────────────┬───────────┬──────────┐
# │ product                  ┆ price_usd ┆ in_stock │
# │ ---                      ┆ ---       ┆ ---      │
# │ str                      ┆ f64       ┆ bool     │
# ╞══════════════════════════╪═══════════╪══════════╡
# │ Google Pixel 7           ┆ 521.9     ┆ true     │
# │ Apple iPhone 14 Pro      ┆ 999.0     ┆ true     │
# │ Samsung Galaxy S23 Ultra ┆ 1199.99   ┆ false    │
# │ OnePlus 11               ┆ 699.0     ┆ true     │
# └──────────────────────────┴───────────┴──────────┘

Footnotes

1. Numpy structured arrays: https://numpy.org/doc/stable/user/basics.rec.html ↩