perf: optimize shuffle array element iteration with pointer arithmetic

Use bulk-append methods with pointer iteration for primitive types
in SparkUnsafeArray instead of per-element get_*() calls which do
multiplication for each index access.

This optimization:
- Adds bulk-append methods for primitive types (i8, i16, i32, i64, f32, f64)
- Uses const generics for compile-time specialization (NULLABLE)
- Avoids repeated multiplication in get_element_offset() calls
- Reads null bitset once per word (64 elements) instead of per-element

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

Author: andygrove