Fix the issue of Optuna tuning mode with LightGBM

[lijm1358]

Fix the issue of Optuna tuning mode with LightGBM, which always fails after LightGBM 4.0.0. verbose_eval has been removed as it did nothing if False was passed; early_stopping_rounds has also been removed and made into a callback.
All tests passed, and also tested with lightgbm==3.3.5, as using the older version of lightgbm was one of the solution in #645.

• before changes:

AutoML directory: test_output
Expected computing time:
Time for tuning with Optuna: len(algorithms) * optuna_time_budget = 21600 seconds
There is no time limit for ML model training after Optuna tuning (total_time_limit parameter is ignored).        
The task is multiclass_classification with evaluation metric logloss
AutoML will use algorithms: ['Random Forest', 'Extra Trees', 'LightGBM', 'Xgboost', 'CatBoost', 'Neural Network']
AutoML will stack models
AutoML will ensemble available models
AutoML steps: ['simple_algorithms', 'default_algorithms', 'ensemble', 'stack', 'ensemble_stacked']
Skip simple_algorithms because no parameters were generated.
* Step default_algorithms will try to check up to 6 models
Optuna optimizes LightGBM with time budget 3600 seconds eval_metric logloss (minimize)
[I 2024-01-08 00:30:52,426] A new study created in memory with name: no-name-e1b40c1d-77c0-4cff-bf8e-78f1677c5906
Exception in LightgbmObjective train() got an unexpected keyword argument 'verbose_eval'
[W 2024-01-08 00:30:52,477] Trial 0 failed with parameters: {'learning_rate': 0.1, 'num_leaves': 1598, 'lambda_l1': 2.840098794801191e-06, 'lambda_l2': 3.0773599420974e-06, 'feature_fraction': 0.8613105322932351, 'bagging_fraction': 0.970697557159987, 'bagging_freq': 7, 'min_data_in_leaf': 36, 'extra_trees': False} because of the following error: The value None could not 
be cast to float..
[W 2024-01-08 00:30:52,478] Trial 0 failed with value None.
Exception in LightgbmObjective train() got an unexpected keyword argument 'verbose_eval'
[W 2024-01-08 00:30:52,581] Trial 1 failed with parameters: {'learning_rate': 0.0125, 'num_leaves': 30, 'lambda_l1': 0.09024841733204539, 'lambda_l2': 0.8785585624049705, 'feature_fraction': 0.5554201923798203, 'bagging_fraction': 0.7307773310574073, 'bagging_freq': 1, 'min_data_in_leaf': 37, 'extra_trees': True} because of the following error: The value None could not be cast to float..
[W 2024-01-08 00:30:52,582] Trial 1 failed with value None.
...

• after changes

[I 2024-01-08 00:32:03,378] A new study created in memory with name: no-name-516d0a9b-d37c-4a2d-811b-856f69d57464
[I 2024-01-08 00:32:03,556] Trial 0 finished with value: 0.1464389479127149 and parameters: {'learning_rate': 0.1, 'num_leaves': 1598, 'lambda_l1': 2.840098794801191e-06, 'lambda_l2': 3.0773599420974e-06, 'feature_fraction': 0.8613105322932351, 'bagging_fraction': 0.970697557159987, 'bagging_freq': 7, 'min_data_in_leaf': 36, 'extra_trees': False}. Best is trial 0 with value: 0.1464389479127149.
[I 2024-01-08 00:32:04,168] Trial 1 finished with value: 0.356323461032784 and parameters: {'learning_rate': 0.0125, 'num_leaves': 30, 'lambda_l1': 0.09024841733204539, 'lambda_l2': 0.8785585624049705, 'feature_fraction': 0.5554201923798203, 'bagging_fraction': 0.7307773310574073, 'bagging_freq': 1, 'min_data_in_leaf': 37, 'extra_trees': True}. Best is trial 0 with value: 0.1464389479127149.
[I 2024-01-08 00:32:05,511] Trial 2 finished with value: 0.11546304145125812 and parameters: {'learning_rate': 0.025, 'num_leaves': 1781, 'lambda_l1': 8.42482357544477e-05, 'lambda_l2': 0.1657023923779856, 'feature_fraction': 0.4006367785978634, 'bagging_fraction': 0.7929826868254444, 'bagging_freq': 5, 'min_data_in_leaf': 22, 'extra_trees': True}. Best is trial 2 with value: 0.11546304145125812.
[I 2024-01-08 00:32:05,804] Trial 3 finished with value: 0.9447620251350354 and parameters: {'learning_rate': 0.0125, 'num_leaves': 1383, 'lambda_l1': 0.0022471032062799873, 'lambda_l2': 
0.0006306544535695586, 'feature_fraction': 0.3303268443196031, 'bagging_fraction': 0.6930031616587092, 'bagging_freq': 3, 'min_data_in_leaf': 51, 'extra_trees': False}. Best is trial 2 with value: 0.11546304145125812.
...

related to: #683, #645

[pplonski]

Thank you @lijm1358!