fix: API-level task/run input and output handling

[monoxgas]

• Replaced Run with RunSummary in the list_runs method of ApiClient to return more aggregated run information.
• Introduced _get_run as a private method in ApiClient to improve encapsulation while maintaining an existing get_run method that now processes run information through the process_run utility function, enhancing data integrity.
• Updated get_run_tasks to now convert tasks into Task objects processed with the context of RawRun, allowing for deeper integration of run context with tasks.
• Modified the get_run_trace logic to enhance how task and trace spans are constructed, introducing better error handling and readability.
• Created a new utility module with functions process_run and process_task, separating concerns and improving the maintainability of the code.
• Added detailed fields with repr=False in various models to improve logging and debugging without exposing sensitive data.

---

Generated Summary:

• Refactored ApiClient methods to use new model structures for runs and tasks.
• Updated list_runs to return RunSummary instead of Run, which simplifies run result handling.
• Introduced helper functions process_run and process_task to transform raw data into structured models while handling references and schemas.
• Updated models to include new RawRun and RawTask classes, separating raw input from processed data.
• Added field representations for tracing attributes to improve debugging without exposing sensitive data in logs.
• Introduced util.py for processing logic, contributing to better code organization and separation of concerns.
• Removed the fast-depends library from pyproject.toml as it is no longer necessary.
• Potential impact includes improved data handling and processing efficiency, reducing the risk of inconsistencies in raw data interpretation.

This summary was generated with ❤️ by rigging