sort-simple-yaml: read and write YAML files as UTF-8

[0xacee]

Problem

sort-simple-yaml opens files with open(filename, 'r+'), so reads and writes use the platform's locale encoding instead of the encoding YAML mandates (UTF-8).

• On a Windows checkout (or any non-UTF-8 locale) a YAML file containing non-ASCII text is decoded as cp1252 — café is read as café — and if a sort is needed the mojibake is written back: silent content corruption.
• On a strict ASCII/POSIX locale the same file raises UnicodeDecodeError.
• Text mode also translates \n to \r\n on write under Windows, so sorting an LF file emits CRLF there but LF on POSIX — platform-dependent output for identical input.

Change

Open the file with encoding='utf-8', newline='' so the hook decodes the encoding YAML specifies and produces byte-identical output on every platform. (newline='' disables \n → \r\n translation on write; rstrip() already normalizes whatever endings were read.)

A non-UTF-8 file now raises a clear UnicodeDecodeError instead of being silently mis-decoded — the file is invalid YAML anyway, since the spec requires UTF-8/16.

Testing

• New test_non_ascii_content_roundtrip writes a UTF-8 file with multi-byte characters and asserts the sorted output preserves them exactly. On the repo's Windows CI job the previous code would fail this (cp1252 decode + \r\n write).
• pytest tests/ — 414 passed, 4 xfailed.

[asottile]

we don't accept ai slop