Part 1 - Core refactor: unify tool/platform APIs and harden launch validation

[eliknebel]

This PR delivers the core refactor for the 1.0.0 release. It unifies the top-level Tool/Platform APIs, restructures
launch validation into explicit stage-based modules, and hardens core correctness around JWT, audience, timestamps,
deployment, registration, state, and nonce validation.

This is the foundation for the follow-on Tool Deep Linking, NRPS, and AGS PRs.

What Changed

• Added unified top-level APIs:
  • Lti_1p3.Tool.login_redirect/2
  • Lti_1p3.Tool.validate_launch/3
  • Lti_1p3.Platform.authorize_redirect/5
• Refactored core validation into explicit ordered validation modules:
  • state
  • registration
  • JWT
  • timestamps
  • deployment
  • nonce
  • message validation
• Normalized launch/auth payloads into typed structs:
  • %Lti_1p3.Tool.Launch{}
  • %Lti_1p3.Platform.AuthorizationPayload{}
• Hardened audience/JWT validation and standardized deterministic error reasons
• Fixed message validator naming/path consistency
• Aligned provider contracts and added provider contract conformance tests
• Cleaned up in-memory provider implementation drift
• Added core telemetry events and integration docs
• Added migration, troubleshooting, and tool/platform integration guides

Why

The existing core had correctness gaps, uneven API ergonomics, and implementation drift between behavior contracts and
runtime code. The service-specific work in later PRs depends on having a stable, spec-correct core to build on.

Reviewer Notes

Focus review on:

• public API shape changes
• validation pipeline correctness
• provider contract consistency
• security-sensitive claim/JWT handling

Service-specific deep linking / NRPS / AGS behavior is intentionally deferred to later PRs in the stack.

Documentation

• README.md
• docs/core_tool_platform_guide.md
• docs/core_migration_guide.md
• docs/core_troubleshooting.md
• docs/telemetry.md
• CHANGELOG.md

[darrensiegel]

It would make sense instead to simply use the harness skills from https://github.com/Simon-Initiative/harness instead of creating and adding specific, new skills here. To do that, you just need to run the $harness-bootstrap skill to create all the necessary artifacts, and then have Codex take a crack at populating them from an informal prompt.

[eliknebel]

It would make sense instead to simply use the harness skills from https://github.com/Simon-Initiative/harness instead of creating and adding specific, new skills here. To do that, you just need to run the $harness-bootstrap skill to create all the necessary artifacts, and then have Codex take a crack at populating them from an informal prompt.

Updated codebase and docs to use AI harness