Extract validateSchema into overridable hook on AbstractSchemaRegistry

[Toluwa Jibodu (djibodu)]

Summary

• Extract the parsedSchema.validate(...) call out of maybeValidateAndNormalizeSchema into a new protected void validateSchema(ParsedSchema, Schema, Config) on AbstractSchemaRegistry. Default implementation preserves the existing IMPORT-mode gate, so behavior is bit-identical for every current caller.
• Same shape as the recent Extract association logic and Extract store query logic into overridable hooks commits on this branch.

Semantics

• parsedSchema.validate(...) has exactly one call site in AbstractSchemaRegistry (the one being moved). All four callers of canonicalizeSchema route through maybeValidateAndNormalizeSchema → the new hook.
• KafkaSchemaRegistry is the only in-repo subclass and doesn't override canonicalizeSchema in a way that touches validation. The maybeValidateAndNormalizeSchema method is private and cannot be overridden, so the hook is purely additive.
• Only subclasses that opt in see different behavior.

Test plan

• AbstractSchemaRegistryTest — new file. Asserts (a) the default hook is invoked during register(...), and (b) a subclass that throws from validateSchema causes registration to fail with the thrown message wrapped in InvalidSchemaException. Documents the extension point and locks the contract.
• Existing core tests — unchanged; no behavior drift expected.
• Local mvn -pl core test hit a CodeArtifact dependency-resolution error (kafka-schema-registry-client:8.0.5-0 cached-miss in third-party-artifacts), so the suite was not run locally. Relying on CI.

🤖 Generated with Claude Code

[confluent-cla-assistant]

🎉 All Contributor License Agreements have been signed. Ready to merge.
_{Please push an empty commit if you would like to re-run the checks to verify CLA status for all contributors.}

[Akhilesh Manjunath (akhileshm1)]

do we still need this given what we discovered yesterday Toluwa Jibodu (@djibodu) ?