ENH: maybe_convert_objects seen NaT speed-up

[BeforeFlight]

I believe after seen.object_ = 1 here:

pandas/pandas/_libs/lib.pyx

Lines 1956 to 1958 in c64c9cb

	if not (convert_datetime or convert_timedelta):
	seen.object_ = 1
	elif util.is_bool_object(val):

should go break as well.

Test:

size = 10**7
arr = list(range(size))
arr[0] = pd.NaT
arr = np.array(arr)

%timeit lib.maybe_convert_objects(arr, convert_datetime=0, convert_timedelta=0)

As now output is:

1.84 s ± 14.3 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)

With break:

57.1 µs ± 887 ns per loop (mean ± std. dev. of 7 runs, 10000 loops each)

And both cases produce same array:

array([NaT, 1, 2, ..., 9999997, 9999998, 9999999], dtype=object)

[jreback]

happy to take a PR that passes tests