feat: Overhaul time-series hyperopt notebook for E2E workflow

This commit completely refactors the `unit_test_synthetic_time_series.ipynb` notebook,

Key improvements include:

- **Data Generation & Preprocessing:**
  - Integrates `generate_synthetic_ts_data` to create a fresh dataset at the start.
  - Adds explicit steps for mean imputation and saving the processed data, ensuring a clean input for the pipeline.

- **Configuration-Driven Workflow:**
  - The notebook now loads `config_hyperopt.yml` to define the model search space (`ts_models`), making it more modular and easier to configure.

- **Robust Objective Function:**
  - The `objective` function is significantly improved with better error handling, including catching `NoFeaturesError` to prevent crashes during the search.
  - It now correctly creates and uses a dedicated experiment directory for each run.
  - The function is parameterized to accept an `outcome_var`, enabling iteration over multiple outcomes.

- **Streamlined Results Processing:**
  - Replaces manual and error-prone log file reading with the `ResultsAggregator` and `MasterPlotter` classes.
  - Adds cells to automatically find the latest experiment results and generate a full suite of analytical plots.
  - Includes data quality checks and visualizations for better insights into the results.

- **Code Cleanup:**
  - Removes numerous commented-out, unused, and fragmented code cells, improving readability and maintainability.

Author: SamoraHunter