fix: replace byte-index string slicing with char-safe truncation

[amitksingh1490]

Summary

Fix byte index N is not a char boundary panics by replacing unsafe byte-index string slicing with character-aware truncation.

Context

The codebase had multiple locations using byte-index string slicing (&str[..N]) which panics when strings contain multi-byte UTF-8 characters (e.g., → = 3 bytes, 🔑 = 4 bytes). This caused runtime crashes when processing tool arguments, error messages, or keys containing non-ASCII characters.

Changes

• forge_domain/auth/new_types.rs: Fixed truncate_key() to use chars().take() and chars().skip() instead of byte indices
• forge_main/tools_display.rs: Fixed error message truncation at byte 77 to use character-safe slicing
• forge_repo/provider_repo.rs: Fixed token debug logging to use character-aware truncation
• forge_domain/result_stream_ext.rs: Fixed test code to use char indices instead of byte indices
• forge_services/fs_patch.rs: Added new tests for from_search_match with multi-byte characters

Key Implementation Details

• Using key.chars().count() for length checks instead of key.len() (byte length)
• Using chars().take(N) for safe prefix truncation
• Using chars().skip(N) for safe suffix extraction
• All fixes prevent panics on emoji, arrows, and other multi-byte UTF-8 characters

Testing

# Run new truncate_key tests
cargo test --package forge_domain -- auth::new_types::tests

# Run fs_patch tests with multi-byte character tests
cargo test --package forge_services -- tool_services::fs_patch::tests

# Run all affected crate tests
cargo test --package forge_domain --package forge_main --package forge_repo

All 5 new tests pass, including:

• Short key handling
• Long ASCII key truncation
• Multi-byte UTF-8 characters (arrows)
• Emoji characters (4-byte UTF-8)
• Exactly 21 char boundary

Links

• Branch: refactor/dotforge-home-directory
• Changes: 5 files, 140 insertions, 6 deletions