Log structured run metadata (configs and metric series) to Neptune using namespaced attribute paths

Steps

1. Create a `Run` from `neptune_scale` with an experiment name
2. Use `run.log_configs({...})` to log single-value hyperparameters/config under a `parameters/` namespace
3. Inside the training loop, call `run.log_metrics(data={...}, step=step)` to log per-step metric series under a `train/` namespace
4. Log datetime values (e.g. training end time) as native Python `datetime` objects via `log_configs`
5. View the organized run in the Neptune app, using namespaces to structure dashboards/reports