(issue #63) add verbosity on training and evaluation procedures throu…

[ChrisNick92]

Add verbosity control to Trainer and Evaluator

Overview

Introduces a verbose parameter to both Trainer and Evaluator, giving users explicit control over console output during training and evaluation. As part of this change, the ConsoleLogger callback has been removed — its logic is now inlined directly into Trainer, eliminating an abstraction that added no real value.

Changes

Trainer

• Added verbose: bool = True parameter. When False, all epoch-level output is suppressed.
• Epoch header [Epoch N/M] and per-epoch metrics (train loss, val loss, time) are now gated by verbose.
• Checkpointer log messages (loss decrease, checkpoint saved) respect verbose via trainer.verbose; error messages always surface.
• EarlyStopper warnings always print regardless of verbose.
• A Training Complete summary block is always printed after training finishes, regardless of verbose:

  ────────────────────────────────────────────────────
   Training Complete
    Best Epoch    : 7
    Train Loss    : 0.234567
    Val. Loss     : 0.123456
    Checkpoint    : checkpoint.pt
  ────────────────────────────────────────────────────

Evaluator

• Added verbose: bool = True parameter. When False, the Reporter output (classification report, confusion matrix, average posteriors) is suppressed. "Evaluation has finished." always prints.

Removed

• callbacks/console_logger.py deleted — logging is now handled inline in Trainer and Evaluator.