refactor: simplify best_fit selection loop with max()

[MAfarrag]

Problem

The p-value selection logic in best_fit uses a manual loop with a sentinel value:

best_name = None
best_pvalue = -1.0
for name, info in fit_results.items():
    pvalue = info[criterion][1]
    if pvalue > best_pvalue:
        best_pvalue = pvalue
        best_name = name
return best_name, fit_results[best_name]

This is 6 lines for what is essentially a max() call. The sentinel -1.0 is a code smell — it works because p-values are always >= 0, but it's not self-evident.

Proposed Solution

Replace with max() using a key function:

best_name = max(
    self._fit_results,
    key=lambda name: getattr(self._fit_results[name], criterion).p_value,
)
return best_name, self._fit_results[best_name]

This is:

• 2 lines instead of 6
• No sentinel value — max() handles empty-check naturally
• More Pythonic — standard pattern for "find the item with the highest X"

Note

This uses getattr on the FitResult dataclass (from #152) where criterion is "ks" or "chisquare", matching the attribute names on FitResult. If #152 is not implemented yet, the dict-based version would be:

best_name = max(
    fit_results,
    key=lambda name: fit_results[name][criterion][1],
)

Scope

• Replace the loop in best_fit with max()
• Verify existing tests still pass (no behavioral change)

Depends On

• refactor: replace untyped nested dicts in fit_all return with FitResult dataclass #152 (FitResult dataclass — for the getattr version)