StringUtils.ToSnakeCase fails for many common scenarios

[rowandh]

StringUtils.ToSnakeCase (used by SnakeCaseNamingStrategy) fails to correctly snake case several common scenarios.

There's plenty more examples when using unicode characters (due to a lack of unicode support), and non-alphanumeric characters, though I'm unsure whether that's by design so have only included examples using ASCII letters and digits.

Steps to reproduce

Add the tests below to Newtonsoft.Json.Tests.Utilities.ToSnakeCaseTest. Observe the output (seen in the comments) is different to expected.

Assert.AreEqual("ab1", StringUtils.ToSnakeCase("AB1")); // FAILS, produces a_b1
Assert.AreEqual("abc1", StringUtils.ToSnakeCase("ABC1")); // FAILS, produces ab_c1
Assert.AreEqual("abc123d_ef456", StringUtils.ToSnakeCase("ABC123DEf456")); // FAILS, produces ab_c123_d_ef456
Assert.AreEqual("abc123def456", StringUtils.ToSnakeCase("ABC123DEF456")); // FAILS, produces ab_c123_de_f456

[YohDeadfall]

@JamesNK I can fix it, but and align it with a policy in System.Text.Json (not merged yet).

[jgoyvaerts]

What's the status of this? StringUtils.ToSnakeCase in it's current version is pretty bad and makes using snake case not a viable option.