Parse throws OverflowException for "0e99999999999" and for representable values near exponent int.MinValue, so ToString("E") output there can't be parsed back #178

What's wrong

In Parse (PreciseNumber/PreciseNumber.cs:1291), exponent = checked(exponent - decimalDigits) subtracts every written fraction digit — including padding zeros — in checked int arithmetic before the value is normalized. The exponent text also goes through int parsing even when the significand is zero.

Reproduced at 907bd2d (invariant culture, NumberStyles.Float):

var x = PreciseNumber.Parse("1e-2147483646", inv);   // fine
var s = x.ToString("E", inv);                         // "1.000000E-2147483646"
PreciseNumber.TryParse(s, NumberStyles.Float, inv, out _); // false — can't read its own output

PreciseNumber.Parse("2.500E-2147483647", inv);  // OverflowException; the value 25e-2147483648 is representable
PreciseNumber.Parse("1.0e-2147483648", inv);    // OverflowException; "1e-2147483648" parses fine
PreciseNumber.Parse("0e99999999999", NumberStyles.Float, inv); // OverflowException; decimal and double return 0

Why it matters

At these magnitudes the plain format would be a ~2 GB string, so "E" is the only text form that can round-trip a value — and it doesn't. Serialized values (e.g. through a round-trip string JSON converter) at the extreme end of the range fail to load. A zero with a large written exponent is valid input that every BCL numeric type accepts.

Suggested fix

• Strip trailing zeros from the collected digits (adjusting decimalDigits) before computing the exponent, or do the exponent arithmetic in long and range-check only after normalizing.
• Return Zero when the significand is zero, regardless of the exponent text (parse the exponent as long/saturating so huge exponents don't throw first).

Acceptance: each input above parses to the expected value, and Parse(x.ToString("E")) round-trips for values with exponents within 10 of int.MinValue/int.MaxValue.