TrimTrailingZeros and AvoidScientificNotation bool variables

[JohnAD]

This is a near-duplicate of #296 that preserves the @vasilbeckk 's work and commits. I will now start making the changes requested in #296.

[JohnAD]

I added the following tests for handling trailing zeros on negative precision. Such as "1000 to the nearest 100 should serialize as 1.0E2"

{"1.0000E3", "1000.0"}, // 1000 to the nearest tenth
{"1.000E3", "1000"},    // 1000 to the nearest one
{"1.0E3", "1.0E3"},     // 1000 to the nearest hundred
{"1E3", "1E3"},         // 1000 to the nearest thousand

(for those not familiar with scientific notation usage, it is required when precision is non-fractional. For example, the population of the U.S. is 314.9 million or, more properly 3.149E8. Any digit after the 9 is an unknown.)

This test does not pass the last two. I'll have to edit some code to fix the bug.

Given that zeros are not always trailing. Perhaps the TrimTrailingZeros option flag should be generically called IgnorePrecision or something like that?

[JohnAD]

Sadly, it appears that covering both cases breaks backwards compatibility. So, I've separated it out. I've added a AvoidScientificNotation option variable. It defaults to true and that default keeps the behavior the same.

As a bonus, I went ahead and made d.ScientificNotationString() a public function and added a unit test.