perf: Make validation check in regressor_column_matrix 300x faster

[MarcoGorelli]

This is something I've noticed on the way towards #2622 (which may not be that far off!)

The builtin Python max would need to iterate over elements, which is much slower than calling the native Series.max (where the algorithm would be in a low-level language)

Example:

In [43]: s = pd.Series(rng.integers(0, 10, size=1_000_000))

In [44]: %timeit max(s) > 1
65.4 ms ± 5.59 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)

In [45]: %timeit s.max() > 1
221 μs ± 16.5 μs per loop (mean ± std. dev. of 7 runs, 10,000 loops each)