Code quality improvements

Description of changes:

• fix template issues that have become compiler errors in Clang 20
• reduce number of Clang-Tidy 20 diagnostics
  • follow-up to !403 (merged)
• adjust codegen jinja templates for pystencils 1.3.7
  • fix missing header includes and compiler warnings about uninitialized block_offset struct members
• expose SIMD vector sizes
  • this will help client codes implement compatible memory-aligned field allocators