Incorrect decoding of 65-bit integer about msgspec

[sigprof]

I suppose that this piece of code does not detect the overflow correctly:

msgspec/msgspec/_core.c

Lines 11902 to 11909 in 2c37da0

	if (
	(digit_count > 19) &&
	(
	(is_float) ||
	((integer_end - integer_start) != 20) ||
	(*integer_start != '1') ||
	(mantissa <= ONE_E18)
	)

Apparently the problem appears with the numbers starting from (1 << 64) + (10 ** 18) + 1, for which the above code will have mantissa = ONE_E18 + 1:

>>> test_value = (1 << 64) + (10 ** 18)
>>> msgspec.json.decode(json.dumps(test_value)), test_value
(19446744073709551616, 19446744073709551616)

>>> test_value = (1 << 64) + (10 ** 18) + 1
>>> msgspec.json.decode(json.dumps(test_value)), test_value
(1000000000000000001, 19446744073709551617)

and continues up to 19999999999999999999 (then at 20000000000000000000 the *integer_start != '1' part triggers, and the code starts behaving correctly again):

>>> test_value = 19999999999999999999
>>> msgspec.json.decode(json.dumps(test_value)), test_value
(1553255926290448383, 19999999999999999999)

>>> test_value = 19999999999999999999 + 1
>>> msgspec.json.decode(json.dumps(test_value)), test_value
(20000000000000000000, 20000000000000000000)