Add missing parameters in OpusMtEnESTransformer's load method

[Irozuku]

Summary

This pull request fixes an issue where loading the model for prediction failed due to schema validation errors. The fix ensures that logging-related parameters defined in the schema are properly handled during model loading, preventing errors when the model is loaded outside of a training context.

---

Type of Change

Check all that apply like this [x]:

• Backend change
• Bug fix
• Frontend change
• CI / Workflow change
• Build / Packaging change
• Documentation

---

Changes (by file)

• opus_mt_en_es_transformer.py: Updated the load method to accept and correctly validate logging configuration parameters (log_train_every_n_epochs, log_train_every_n_steps, log_validation_every_n_epochs, log_validation_every_n_steps) defined in the schema, avoiding validation errors when loading the model for prediction.

---

Testing (optional)

• Verify that training behavior remains unchanged.
• Confirm that schema validation no longer raises errors for logging-related parameters.
• Verify tha the prediction is working properly with the trained model.