feat: Cache-aligned bytecode storage

[zharinov]

Summary

• Add AlignedVec type using Vec<Block> trick for 64-byte alignment
• Add include_query_aligned! macro for zero-copy embedded bytecode
• Update ByteStorage enum with Static and Aligned variants
• New Module API: from_aligned, from_static, from_path, load
• Deprecate Module::from_bytes

Why

Bytecode sections are 64-byte aligned internally. For DFA deserialization and cache efficiency, the buffer itself must start at a 64-byte boundary. Standard Vec<u8> provides no alignment guarantees.