fix: skip UTF-8 BOM when processing front matter

[StressTestor]

Disclosure: this PR was written by an AI agent (Claude Code) acting on the user's behalf, not the user personally.

what

yq --front-matter=process drops the opening --- when the input file starts with a UTF-8 BOM, which loses data:

$ printf '\xEF\xBB\xBF---\ntitle: Test\n---\n' > test.md
$ yq --front-matter=process test.md
title: Test
---

The opening --- is gone. This is #2496.

why

front_matter.go's Split() wraps the file in a bufio.Reader without skipping a leading BOM. The first line read is then \xEF\xBB\xBF---, so the --- separator is not recognised and that line is treated as content instead of the front-matter delimiter.

fix

Skip the BOM with utfbom.Skip before wrapping the reader, the same way decoder_csv_object.go already does. utfbom is already a direct dependency.

cleanReader, _ := utfbom.Skip(file)
reader = bufio.NewReader(cleanReader)

Applied to both the file and stdin paths. utfbom.Skip replays non-BOM bytes unchanged, so files without a BOM are byte-identical.

tests

TestFrontMatterSplitWithBOM in front_matter_test.go feeds a BOM-prefixed document and asserts the front matter and content split correctly. It fails before the fix (the front matter starts with the BOM) and passes after.

Fixes #2496.

[mikefarah]

@cursoragent review this

[ccoVeille]

I find strange to add such dependency to skip UTF BOM.

The repository is stale (which is OK because of the feature stability), but as you use only Skip method.

Is there a need in yq to handle all UTF BOM. Isn't UTF-8 enough?

var utf8BOM = []byte{0xEF, 0xBB, 0xBF}

// StripUTF8BOM returns a reader that skips a leading UTF-8 BOM if present.
func StripUTF8BOM(r io.Reader) io.Reader {
	br := bufio.NewReader(r)

	peek, err := br.Peek(3)
	if err == nil && bytes.Equal(peek, utf8BOM) {
		_, _ = br.Discard(3)
	}

	return br
}

[ccoVeille]

What do you think @mikefarah