update and expand scientific docstrings for optimizer, scheduler, trainer

- Expanded module-level and function/class docstrings in optimizer.py, scheduler.py, and trainer.py
- Described mathematical foundations, theoretical motivations, and provided detailed usage examples for students
- All docstrings in Russian, clear scientific style

test(training): add comprehensive tests for optimizer, scheduler, and trainer modules

- Added new test files for get_optimizer, get_linear_schedule_with_warmup, and Trainer
- Tests cover parameter handling, edge cases, and expected learning dynamics (lr schedules and loss behavior)
- Trainer now logs average epoch losses to self.loss_history for testability and analysis

refactor(training/trainer): log epoch loss to loss_history for downstream analysis and tests

BREAKING CHANGE: Trainer.loss_history is a new attribute consolidating average losses per epoch, enabling robust learning dynamics assertions in tests

llm/tests/training/test_optimizer.py

@@ -0,0 +1,35 @@
+import pytest
+import torch.nn as nn
+from llm.training.optimizer import get_optimizer
+
+class DummyModel(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.linear = nn.Linear(10, 1)
+
+def test_get_optimizer_adamw():
+    model = DummyModel()
+    optimizer = get_optimizer(model, lr=1e-3, weight_decay=0.02, optimizer_type="adamw")
+    assert optimizer.__class__.__name__ == 'AdamW'
+    assert optimizer.defaults['lr'] == 1e-3
+    assert optimizer.defaults['weight_decay'] == 0.02
+
+def test_get_optimizer_adam():
+    model = DummyModel()
+    optimizer = get_optimizer(model, lr=1e-4, weight_decay=0.01, optimizer_type="adam")
+    assert optimizer.__class__.__name__ == 'Adam'
+    assert optimizer.defaults['lr'] == 1e-4
+    assert optimizer.defaults['weight_decay'] == 0.01
+
+def test_get_optimizer_sgd():
+    model = DummyModel()
+    optimizer = get_optimizer(model, lr=0.1, optimizer_type="sgd")
+    assert optimizer.__class__.__name__ == 'SGD'
+    assert optimizer.defaults['lr'] == 0.1
+    # SGD: weight_decay по умолчанию 0 для этого вызова
+    assert optimizer.defaults['momentum'] == 0.9
+
+def test_get_optimizer_invalid():
+    model = DummyModel()
+    with pytest.raises(ValueError):
+        get_optimizer(model, optimizer_type="nonexistent")