Use the production allocator in PyPI type benchmarks (#21915)
The `uv_pypi_types` benchmarks use the system allocator, allowing glibc
allocation variance to dominate small digest measurements. Explicitly
link `uv_performance_memory_allocator`, as the `uv` and workspace
benchmarks do, so these measurements use the allocator selected for uv.
Co-authored-by: zaniebot <242828183+zaniebot@users.noreply.github.com>