The exported result is incorrect if the document had not UTF-8 encoding

[yhatt]

Version of Marp Tool

Marp for VS Code v3.1.0

Operating System

macOS

Environment

OS version: macOS Sequoia 15.3.1
VS Code version (Marp for VS Code): VS Code 1.97.2

How to reproduce

1. Create a new Markdown file just like this:

🐣

2. Save Markdown file with UTF-16 LE encoding
3. Export slide deck to any format.

Expected behavior

Show a slide page with the content 🐣

Actual behavior

A slide page has incorrect encoding.

Additional information

Originally reported in marp-team/marp#558.

Extension could not deal with that due to lack of VS Code API for getting a encoding for the current document. But now the latest VS Code 1.98 has a newly proposed API for getting encoding.
https://code.visualstudio.com/updates/v1_98#_text-encodings
https://github.com/microsoft/vscode/blob/501ee833b16b8e83ba656c46e0888aadd9d2db04/src/vscode-dts/vscode.proposed.textDocumentEncoding.d.ts#L1

To fix that, Marp for VS Code should use an temporary file as an input if detected non UTF-8 encoding.

marp-vscode/src/marp-cli.ts

Line 30 in bc09bd5

if (doc.uri.scheme === 'file' && !doc.isDirty) {