MNE-BIDS should not include a BOM when writing TSV files

[scott-huberty]

From #1528 .. in #625 MNE-BIDS started writing text files with encoding "utf-8-sig". However the BIDS spec only requires 'utf-8' for tabular files:

Haha funny, so I initially didn't want to have a BOM but then we decided otherwise :)

I did a bit of digging online and it seems that NOT having the BOM is the "correct" way of doing things. And newer versions of Excel (at least in M365) read non-BOM'd TSV files correctly! In fact, a quick test shows actually that the situation has sort of reversed! – not having a BOM is BETTER for Excel, at least when reading TSVs, as can be seen here:

# test.py
#
# /// script
# requires-python = ">=3.14"
# dependencies = [
#     "pandas>=3.0.1",
# ]
# ///
import pandas as pd

data = pd.DataFrame.from_dict({
    "Foo": [1, 2, 3],
    "Bar": [4, 5, 6],
    "Baz": [7, 8, 9]
})

data.to_csv("data.tsv", index=False, encoding="utf-8", sep="\t")
data.to_csv("data-bom.tsv", index=False, encoding="utf-8-sig", sep="\t")

Run via:

uv run test.py

No BOM – loads correctly upon a double-click on the file:

With BOM – does not load correctly:

So I would say we drop the BOM. How to handle this potentially breaking change (then again, who would really be affected?) – I leave up to @sappelhoff to decide 😃

Originally posted by @hoechenberger in #1528 (comment)

[sappelhoff]

As mentioned in the comment thread, I think we can drop the BOM and if we can handle reading in a backwards compatible manner, then that'd be great.