refactor: Optimize NullMath utilities

[DaveSkender]

Some targeted performance optimizations for NullMath utility functions and more specific unit tests.

[DaveSkender]

BEFORE

Method	Mean	Error	StdDev	Median
AbsDblVal	0.9209 ns	0.1937 ns	0.0106 ns	0.9176 ns
AbsDblNul	0.9608 ns	0.4442 ns	0.0243 ns	0.9579 ns
RoundDecVal	9.2071 ns	0.5013 ns	0.0275 ns	9.2205 ns
RoundDecNul	0.7303 ns	0.1966 ns	0.0108 ns	0.7342 ns
RoundDblVal	1.9981 ns	0.2965 ns	0.0163 ns	2.0021 ns
RoundDblNul	0.6847 ns	0.7638 ns	0.0419 ns	0.6612 ns
Null2NaNDecVal	1.4518 ns	0.6174 ns	0.0338 ns	1.4352 ns
Null2NaNDecNul	0.7991 ns	0.2103 ns	0.0115 ns	0.7930 ns
Null2NaNDblVal	0.0043 ns	0.0754 ns	0.0041 ns	0.0047 ns
Null2NaNDblNul	0.0110 ns	0.2700 ns	0.0148 ns	0.0052 ns
NaN2NullDblVal	1.0894 ns	0.0658 ns	0.0036 ns	1.0880 ns
NaN2NullDblNul	0.6967 ns	0.1059 ns	0.0058 ns	0.6987 ns
NaN2NullNaNVal	0.0441 ns	0.1247 ns	0.0068 ns	0.0416 ns
NaN2NullNanNul	1.2751 ns	0.1600 ns	0.0088 ns	1.2734 ns

AFTER

Method	Mean	Error	StdDev	Median
AbsDblVal	0.2521 ns	0.3440 ns	0.0189 ns	0.2603 ns
AbsDblNul	0.0475 ns	0.1109 ns	0.0061 ns	0.0458 ns
RoundDecVal	5.4814 ns	0.6534 ns	0.0358 ns	5.5017 ns
RoundDecNul	2.7583 ns	0.3855 ns	0.0211 ns	2.7522 ns
RoundDblVal	1.1781 ns	0.1948 ns	0.0107 ns	1.1728 ns
RoundDblNul	0.0344 ns	0.2516 ns	0.0138 ns	0.0327 ns
Null2NaNDecVal	1.4521 ns	0.6041 ns	0.0331 ns	1.4596 ns
Null2NaNDecNul	0.0000 ns	0.0000 ns	0.0000 ns	0.0000 ns
Null2NaNDblVal	0.0001 ns	0.0035 ns	0.0002 ns	0.0000 ns
Null2NaNDblNul	0.0014 ns	0.0442 ns	0.0024 ns	0.0000 ns
NaN2NullDblVal	0.0682 ns	0.1296 ns	0.0071 ns	0.0666 ns
NaN2NullDblNul	0.0544 ns	0.1359 ns	0.0075 ns	0.0560 ns
NaN2NullNaNVal	0.0383 ns	0.1076 ns	0.0059 ns	0.0408 ns
NaN2NullNanNul	0.0607 ns	0.1026 ns	0.0056 ns	0.0592 ns