作用：在一张极小的手工簇图上运行着丝粒流程的并查集合并步骤 build_supercluster_unionfind.py。
入口：run_demo.sh
输入：input/clusters.list（8 个重复簇，每行一个）与 input/pairs.jaccard_ge0.7.tsv（6 条不低于 0.7 截断的 Jaccard 对，制表符分隔 A B Jaccard：c1-c2 0.91、c1-c3 0.88、c2-c3 0.95、c4-c5 0.77、c6-c7 0.83、c7-c8 0.81）。
输出：work/cluster2super.tsv 与 work/super_summary.tsv；参考结果在 expected_output/，内容为 super_0001 3 c1,c2,c3、super_0002 2 c4,c5、super_0003 3 c6,c7,c8。
运行：bash run_demo.sh（把输入复制进 work/ 后在其中运行该步骤，并打印两个输出文件）。
工具：Python 3 标准库，无第三方依赖；运行时间远低于 1 秒。
English
**Purpose**: Run the union-find merging step of the centromere pipeline (`build_supercluster_unionfind.py`) on a tiny hand-made cluster graph.
**Entry point**: `run_demo.sh`
**Inputs**: `input/clusters.list` (8 repeat clusters, one per line) and `input/pairs.jaccard_ge0.7.tsv` (6 pairwise Jaccard indices at or above the 0.7 cutoff, tab-separated `A B Jaccard`: c1-c2 0.91, c1-c3 0.88, c2-c3 0.95, c4-c5 0.77, c6-c7 0.83, c7-c8 0.81).
**Outputs**: `work/cluster2super.tsv` and `work/super_summary.tsv`; reference copies are in `expected_output/` and read `super_0001 3 c1,c2,c3`, `super_0002 2 c4,c5`, `super_0003 3 c6,c7,c8`.
**Run**: `bash run_demo.sh` (copies the inputs into `work/`, runs the step there and prints both output files).
**Tools**: Python 3 standard library, no third-party dependency; runtime is well under a second.