Type Safety API - remove json.dumps()

[bordumb]

Manual json.dumps() scattered throughout the codebase is error-prone.

1. Pydantic models with auto-serialization (simplest)
   - Define InvestigationOutcome, JobResult as Pydantic models
   - Repository methods accept models, handle model.model_dump_json() internally
   - Type-safe, IDE catches errors

And do this elsewhere

[bordumb]

Look into this as well:

✦ Using json.dumps(context.model_dump(mode="json")) is a correct and robust solution, but for a modern Pydantic V2 codebase (which
dataing appears to be, given the usage of ConfigDict), the absolute best solution—preferred by top-tier engineering teams for
performance and idiom—is to use context.model_dump_json().

Here is why context.model_dump_json() is superior:

1. Performance: It performs serialization in a single pass (often in optimized Rust code via pydantic-core), avoiding the
   overhead of creating a massive intermediate Python dictionary just to immediately serialize it to a string.
2. Memory Efficiency: It reduces memory pressure by not allocating Python objects for every field in your context before
   serialization.
3. Idiomatic: It is the dedicated method provided by the framework for this exact purpose.

[bordumb]

From Gemini:

✦ I have identified several areas where json.dumps() is used. Replacing json.dumps() with Pydantic's model_dump_json() or a custom
encoder that handles complex types (like date, datetime, UUID) would be beneficial for robustness, especially in:

1. dataing/src/dataing/adapters/db/app_db.py: Numerous instances of json.dumps() are used to serialize metadata, events,
   finding, checkpoint, changes, context, etc., before inserting them into the database. If these dictionaries contain datetime
   objects or other non-serializable types, it will cause runtime errors similar to the one I just fixed.
2. dataing/src/dataing/entrypoints/worker.py: Used for serializing status updates and errors. While currently simpler, using a
   robust serializer would prevent future regressions if complex objects are added to the payloads.
3. dataing/src/dataing/entrypoints/api/routes/investigations.py and notifications.py: Used for SSE data. Ensuring robust
   serialization here is good practice.
4. dataing/src/dataing/adapters/notifications/webhook.py: Used for webhook bodies.

Recommendation:
It is highly recommended to refactor these usages, particularly in app_db.py, to use a robust JSON serializer. A helper function
like json_dumps that uses Pydantic's TypeAdapter or a custom encoder would be the best approach to ensure consistency and safety
across the codebase.