feat: add progress callback to run_grid and Study.run

[jc-macdonald]

Summary

Adds a progress callback parameter to run_grid() and Study.run() so users can monitor long-running sweeps (#77).

Changes

• New callback: ProgressCallback | None = None parameter on run_grid()
  • Called after each trial with (trial_index, total_trials, trial_result)
  • In serial mode (n_jobs=1), fires immediately after each trial
  • In parallel mode, fires post-hoc after all trials complete
• Study.run() forwards callback to all run_grid() calls
• ProgressCallback type alias: Callable[[int, int, TrialResult], None]

Usage

from tqdm import tqdm
bar = tqdm(total=len(grid))
run_grid(..., callback=lambda i, n, r: bar.update(1))

Tests

• 4 new tests (2 in test_runner.py, 2 in test_study.py)
• 225 total tests passing, 99.53% coverage

Closes #77