fix: reject invalid SemVer identifiers when parsing

Per SemVer 2.0.0 the core version numbers and numeric prerelease
identifiers must not include leading zeroes, and prerelease/build-metadata
identifiers must be composed only of ASCII alphanumerics and hyphens.

Add IsValidIdentifiers using a vectorized SearchValues lookup for the
allowed character set and ulong.TryParse to detect numeric identifiers
for the leading-zero check. Add failing-then-passing tests for both
invalid and still-valid inputs.

CHANGELOG.md

@@ -16,6 +16,10 @@ This project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.htm
 
 ### Changed
 
+### Fixed
+
+- Reject invalid SemVer identifiers when parsing versions.
+
 ### Removed
 
 [1.0.0]: https://code.geekeey.de/geekeey/semver/releases/tag/1.0.0