refactor: address claims accessor review follow-ups

[SeanROlszewski]

Follow-up to the four review comments on #490:

1. Trim the claims() doc comment — reduced to the summary line.
2. Arc in claims() -> Vec<Arc<FieldElement>> — investigated and it's load-bearing: uniffi objects only implement Lower behind an Arc, so sequences of objects must be Vec<Arc<T>>. Dropping it fails scaffolding compilation with the trait bound 'FieldElement: uniffi::Lower<UniFfiTag>' is not satisfied (bare owned objects are accepted only as top-level return values, which is why sub() gets away without one). Left a code comment so the next reader doesn't retry it.
3. Remove claims_hex() — gone; callers now read claims() and encode the specific elements they need, making per-claim selection the deliberate act.
4. Trim the fetch_credential doc comment — reduced to the summary + # Errors (kept for missing_errors_doc).

Testing: cargo test -p walletkit-core --lib green (tests updated for the removed method); clippy (pedantic) clean.

---

Note

Low Risk
API surface and documentation cleanup only; behavior of claim access is unchanged aside from removing the convenience hex helper.

Overview
Follow-up on credential claims API review: claims_hex() is removed so FFI callers use claims() and encode only the field elements they need (e.g. to_hex_string()), instead of a bulk hex helper.

claims() keeps returning Vec<Arc<FieldElement>>; the long disclosure-oriented rustdoc is replaced with a short summary plus an inline note that Arc is required for UniFFI sequences of objects. fetch_credential rustdoc is shortened (no tie-in to proof generation / claims_hex). Tests are updated accordingly.

^{Reviewed by Cursor Bugbot for commit 700c632. Bugbot is set up for automated code reviews on this repo. Configure here.}

[paolodamico]

no need to have this. removing the Arc will immediately raise a compilation error to anyone confused like me