Escape C1 control characters with Unicode escapes

[mnnekrashevich]

C1 control characters (U+0080–U+009F) were escaped using \xNN.
Because \xNN represents a raw byte, converting valid UTF-8 strings with
Macro.to_string/1 could change their encoding.

For example, U+0090 (<<0xC2, 0x90>>) was rendered as "\x90" and evaluated
back to the invalid UTF-8 byte <<0x90>>.

This change uses \u00NN for C1 control characters while retaining \xNN
for ASCII control bytes. It also adds a regression test covering the complete
U+0080–U+009F range.

Tests:

• make format
• bin/elixir lib/elixir/test/elixir/macro_test.exs
• make test_stdlib
• bin/elixir lib/elixir/test/elixir/system_test.exs

Assisted-by: Codex:GPT-5

[Inflyser]

It's divine! You nailed it, bro!

[josevalim]

💚 💙 💜 💛 ❤️