Avoid full-string scans in token-budget truncation (#49712)

## Why

Token-budget truncation uses byte counts to estimate tokens, but previously scanned the entire string to count characters, including the discarded middle.

## What changed

Use UTF-8 boundary lookups to select the retained prefix and suffix for token-budget truncation. Skip character counting for a zero-token budget and defer it to paths that need character counts.

## Testing

Add regression coverage for multibyte characters crossing both retained boundaries, checking the truncated output and original token estimate.

GitOrigin-RevId: 37fee0c8d4ede8e74b491d3d641b0ea0e9003dd4

codex-rs/utils/string/src/truncate/tests.rs

@@ -109,6 +109,20 @@ fn truncate_middle_tokens_handles_utf8_content() {
     assert_eq!(tokens, Some(16));
 }
 
+#[test]
+fn truncate_middle_tokens_keeps_partial_characters_out_of_both_ends() {
+    for (input, expected) in [
+        ("a😀b😀c", "a…2 tokens truncated…c"),
+        ("é中😀é", "é…2 tokens truncated…é"),
+        ("😀😀😀", "…2 tokens truncated…"),
+    ] {
+        assert_eq!(
+            truncate_middle_with_token_budget(input, /*max_tokens*/ 1),
+            (expected.to_string(), Some(3)),
+        );
+    }
+}
+
 #[test]
 fn truncate_middle_bytes_handles_utf8_content() {
     let s = "😀😀😀😀😀😀😀😀😀😀\nsecond line with text\n";