fix(int128): Use all 16 decimal digits to represent int128 min/max in double number correctly

[TreeHunter9]

The value I128_MIN_dbl was not represented precisely, as it didn't utilize all 16 decimal digits (52 bits of mantissa) available in a double-precision number. As a result, the value was rounded down to the closest representable number in the double format.

This inaccuracy led to an error in the following query:

SELECT CAST(-POWER(2, 127) AS INT128) FROM RDB$DATABASE;
=======================
Statement failed, SQLSTATE = 22003
arithmetic exception, numeric overflow, or string truncation
-numeric value is out of range

[AlexPeshkoff]

Why whitespace?

[TreeHunter9]

To look in line with I128_MIN_dcft, like it's done with I128_MIN_dbl and I128_MAX_dbl