write_excel(..., autofit=True) auto-fits to source, not formatted data

[mcrumiller]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

import polars as pl

df = pl.DataFrame({
    "Text": ["Hi there", "longer sentence", "this is left aligned"],
    "Number": [1.0, 2.345, 0.398374937341998375913758438431441],
})

df.write_excel("out.xlsx", autofit=True)

Issue description

When using autofit, it appears the column width is calculated from the original number, not from the formatted result. In the example above, the column width is as it would be had we kept many decimals in the output format, and as a result, we have unused whitespace.

Expected behavior

Should fit to truncated, output-formatted decimal.

Installed versions

main

[alexander-beedie]

FYI: the behaviour of autofit is an XlsxWriter feature rather than a Polars feature - we just call "worksheet.autofit()" if it is requested. So, unless we want to write a custom version of that ourselves (and/or contribute enhancements upstream?), you'll need to make a feature request for improvements over there ;)

Note that "better autofit" is already mentioned in the XlsxWriter roadmap:
jmcnamara/XlsxWriter#1028

[mcrumiller]

Thanks @alexander-beedie, definitely not polars fault!