Speed up framing, linting, and compilation

[jviotti]

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

[cubic-dev-ai]

No issues found across 29 files

[augmentcode]

🤖 Augment PR Summary

Summary: This PR updates core dependencies and refactors schema framing/linting/compilation paths to reduce unnecessary work and enable compiling/validating from a specific subschema entrypoint.

Changes:

• Bumps vendored dependency SHAs (core/blaze/jsonbinpack/codegen/jsonschema) and updates jsonschema to 14.7.0.
• Adds an explicit entrypoint concept to the Blaze compiler API and introduces CompilerInvalidEntryPoint for clearer failures.
• Refactors Blaze compilation to compute reachability starting from the chosen entrypoint and to always bundle in the high-level compile path.
• Reworks SchemaFrame reachability into a cached, graph-based traversal keyed by base location (intended to speed repeated reachability queries).
• Updates transformer callbacks to include a boolean indicating “mutable on check” / “applied on apply”, and wires this through generators/tools.
• Improves lint auto-fix UX (progress output) and adds lint --format/--keep-ordering options.
• Extends CLI compile and validate commands with --entrypoint (compile/validate a subschema by pointer/URI).

Technical Notes: Entry-point compilation uses a pre-analysed SchemaFrame to avoid redundant bundling/framing where possible, and linter rules reuse an Evaluator instance to reduce per-check overhead.

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

[augmentcode]

Review completed. 1 suggestions posted.

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

[augmentcode]

If entrypoint is exactly "#", pointer_string becomes empty and this returns <root>#, which likely won’t match frame.root() (no fragment) and makes the root-pointer form unusable. Consider special-casing "#" to return frame.root() directly.

Severity: medium

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