realizar 0.6.8

Pure Rust ML inference engine built from scratch - model serving for GGUF and safetensors
Documentation
# <type>(<scope>): <subject>
#
# <body>
#
# <footer>
#
# Types:
#   feat     - New feature
#   fix      - Bug fix
#   perf     - Performance improvement
#   refactor - Code restructuring (no feature/fix)
#   test     - Add/update tests
#   docs     - Documentation only
#   style    - Formatting, no code change
#   chore    - Maintenance (deps, config)
#   ci       - CI/CD changes
#
# Scope examples: parser, transformer, quantize, tokenizer, api, cli
#
# Subject:
#   - Imperative mood ("add" not "added")
#   - No period at end
#   - Max 50 characters
#
# Body:
#   - Wrap at 72 characters
#   - Explain WHAT and WHY, not HOW
#   - Reference issues: Refs #123
#
# Footer:
#   - Breaking changes: BREAKING CHANGE: <description>
#   - Issue references: Closes #123, Fixes #456
#
# Example:
#
# feat(parser): add GGUF v3 header parsing
#
# Implement parser for GGUF version 3 header format including
# new alignment field. Required for llama.cpp 0.3+ compatibility.
#
# - Parse new alignment metadata field
# - Validate version >= 3 for new fields
# - Backward compatible with v2 files
#
# Refs #234
# Closes #567