Add count column to markdown results summary

[Bortlesboat]

Summary

• add per-task
  _samples to the exported evaluation results dictionary
• render a new Count column in make_results_table markdown output
• add unit tests covering populated and missing count behavior

Why

Issue #804 asks for the markdown summary to show how many questions were evaluated so low scores are easier to interpret.

Validation

• python3 -m py_compile src/lighteval/logging/evaluation_tracker.py src/lighteval/utils/utils.py tests/unit/utils/test_utils.py
• Full unit test run was not possible in this environment because WSL Python is missing pip/pytest.