Upgrade Core to fe450b982907f99e542a0cfc78bc60d2b600ff7a

[jviotti]

Signed-off-by: Juan Cruz Viotti jv@jviotti.com

[augmentcode]

🤖 Augment PR Summary

Summary: This PR bumps the vendored sourcemeta/core dependency and updates JSON BinPack to the new Core APIs.

Changes:

• Updated the Core pinned revision in DEPENDENCIES to fe450b98….
• Adjusted schema-transformer callbacks to match Core’s updated callback signature.
• Reworked JSON Pointer hashing to better combine multi-word property hashes.
• Optimized SchemaFrame analysis by switching several caches from std::map to std::unordered_map and caching the standalone() result.
• Improved reachability analysis for dynamic anchors by indexing dynamic anchor locations by fragment.
• Changed the JSON Schema wrap() API to accept a pre-analysed SchemaFrame + Location and to return additional “base” pointer information.

Technical Notes: The Core upgrade changes callback semantics (mutability/applied flag) and relies more heavily on WeakPointer hashing behavior.

_{🤖 Was this summary useful? React with 👍 or 👎}

[augmentcode]

Review completed. 2 suggestions posted.

Comment augment review to trigger a new review at any time.

[augmentcode]

SchemaTransformer::apply() can still call the callback for the final non-mutating rules pass (with the boolean argument set to false), so assert(applied) could trip in debug builds even though this is meant to be a no-op callback.

Severity: medium

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}

[augmentcode]

This reinterpret_cast<const std::uint64_t *>(&hash.a) reads a __uint128_t through a uint64_t*, which can violate strict-aliasing/alignment and is undefined behavior under optimization. Since this feeds WeakPointer hashing for unordered_* containers, any UB here can cause very hard-to-debug misbehavior.

Severity: high

_{🤖 Was this useful? React with 👍 or 👎, or 🚀 if it prevented an incident/outage.}