Add context.readResource() to MethodContext for user extension models

Problem

User extension models that need to read back previously written data (e.g., reading discovery results during a pair operation) must use context.dataRepository.getContent() directly. This API uses positional arguments with an internal ModelType object:

context.dataRepository.getContent(context.modelType, context.modelId, dataName)

This is brittle and asymmetric — context.writeResource(specName, name, data) provides a clean high-level API for writes, but there is no equivalent for reads. Extension authors must:

1. Know to use context.modelType (a ModelType object), not context.definition.id (a string) — the error when you get this wrong (type.toDirectoryPath is not a function) gives no hint about the actual problem.
2. Manually decode the returned Uint8Array and parse JSON.
3. Understand internal repository semantics that writeResource otherwise abstracts away.

Proposed Solution

Add a context.readResource(dataName) method to MethodContext that mirrors writeResource:

readResource(dataName: string): Promise<Record<string, unknown> | null>

This would:

• Use context.modelType and context.modelId internally
• Decode the Uint8Array and parse JSON automatically
• Return null if the data doesn't exist

Summary

Changes would be needed in:

• MethodContext interface in src/domain/models/model.ts to add the readResource method signature
• method_execution_service.ts to provide the implementation when building the context
• Symmetry with writeResource — same level of abstraction for both reads and writes

This keeps extension models working at the resource abstraction level without reaching into internal repository APIs.

Alternatives

• Document the dataRepository.getContent() pattern more prominently (still brittle)
• Use a named-object argument style for getContent instead of positional args (helps but still exposes internals)