perf: parallelize computationally expensive block processing in pkgs/types/src/block.zig

## Background

Raised as a follow-up suggestion by @anshalshukla in a review comment on #686 (devnet4):

> as a followup PR we can do parallel processing here something similar to how it can be done using rayon, this being a computationally expensive task.

See: https://github.com/blockblaz/zeam/pull/686#discussion_r3056196584

## Problem

Block processing in `pkgs/types/src/block.zig` is computationally expensive and currently runs serially. This is a bottleneck that could benefit from parallel execution.

## Proposed Solution

Introduce parallel processing for the expensive computation in `block.zig`, analogous to how [Rayon](https://docs.rs/rayon) enables data-parallelism in Rust. Options in Zig include:

- Using `std.Thread.Pool` / `std.Thread.spawn` for coarse-grained parallelism
- Splitting work items into chunks and processing in parallel worker threads

## Notes

- This is a performance improvement, not a correctness fix — low priority
- Needs benchmarking before and after to confirm wins
- Should not block devnet4 or any current PRs

Start by reading pkgs/types/src/block.zig and identify the computationally expensive block-processing work that currently runs serially. Establish a benchmark before changing execution, then evaluate std.Thread.Pool or std.Thread.spawn and chunking options. Done means parallel processing is correct and benchmarks demonstrate a worthwhile improvement without blocking current work.

Written by the indexing model from the issue text.