test: fuzz metadata_limits validators with multi-byte UTF-8 sequences (closes #613)

[nuhumusamagaji]

Closes #613

---

Summary

Addresses #613 by hardening metadata_limits validators for multi-byte UTF-8 input and adding property-based test coverage.

• Length policy documented and fixed: All string validators now count Unicode scalar values (.chars()), not UTF-8 byte length (String::len()). A 4-byte glyph like 😀 counts as one character toward the limit.
• Control-character rejection: Validators reject Unicode control characters (char::is_control()) with Error::InvalidInput.
• No unwrap() in validators: Vector iteration uses get(i).ok_or(Error::InvalidInput)?.
• Proptest fuzzers: 8 property tests (256 cases each) for validate_question_length, validate_outcome_length, validate_category_metadata, and validate_tag_metadata, using proptest::string::string_regex for printable ASCII and supplementary-plane symbols.
• Deterministic edge cases: Empty/at-limit/just-over-limit behavior and explicit 4-byte glyph (U+1F600) tests.

Files changed

File	Change
contracts/predictify-hybrid/src/metadata_limits.rs	scan_metadata_text helper, scalar-length checks, control-char rejection, removed unwrap()
contracts/predictify-hybrid/src/metadata_limits_tests.rs	Unit edge cases + proptest_fuzz module

Length semantics (documented in module rustdoc)

Measure	Used?	Notes
Unicode scalar values	✅	Matches docs ("500 characters") and CreationValidator
UTF-8 byte length	❌	Previous behavior; inconsistent with docs
Control characters	❌ rejected	Returns InvalidInput

Test plan