Evaluate replacing custom OpenAPI parser with established library

[petersalomonsen]

Context

We currently use a custom OpenAPI parser implementation to generate TypeScript types and Zod schemas from the NEAR OpenAPI specification. Recent issues (like duplicate schema generation) raise the question of whether we should use an established library instead.

Current Issues

• Maintenance burden when OpenAPI spec changes
• Edge cases like duplicate schema names require manual fixes
• Limited support for complex OpenAPI features (oneOf, anyOf, allOf)

Documentation

See OPENAPI_PARSER_EVALUATION.md for detailed analysis of pros/cons.

Recommendation

For now, keep the custom implementation but improve error handling and test coverage. Consider a hybrid approach in the future if maintenance becomes problematic.

Related Context

• Recent fix for duplicate GenesisConfigRequest schema generation
• NEAR-specific requirements: snake_case to camelCase conversion
• Need for both TypeScript types and Zod schemas for runtime validation