Add runnable `.sql` examples and regression tests for the built-in graph algorithms

### What & why

GeaFlow ships 19 built-in graph algorithms (PageRank, SSSP, WCC, Louvain, LPA,
KCore, TriangleCount, JaccardSimilarity, …), but the `geaflow-dsl-runtime`
regression suite only covers some of them. Filling in end-to-end
`CALL ... YIELD` tests for the rest guards against future refactors quietly
breaking an algorithm. Great if you want to *understand* what graph algorithms
actually compute.

GeaFlow 有 19 个内置图算法,但回归测试只覆盖了一部分。给其余算法补端到端的
`CALL ... YIELD` 测试,能防止后续重构悄悄改坏算法。想真正搞懂"图算法到底算了什么"的同学,这题很合适。

### The task

Pick a few algorithms that have no tests (or thin coverage) and add `.sql`
queries + `.txt` expected outputs.

挑几个尚无测试或覆盖薄弱的算法,补 `.sql` 查询 + `.txt` 期望输出。

### Where to look

1. Test graph definition: `geaflow-dsl-runtime/src/test/resources/modern_graph.sql`.
2. Query template: existing `query/gql_algorithm_001.sql`
(`CALL SSSP(1) YIELD (vid, distance)` style); expected output at `expect/gql_algorithm_001.txt`.
3. Test entry point: `geaflow-dsl-runtime/.../runtime/query/GQLAlgorithmTest.java`.

### Done when

- [ ] At least one `.sql`/`.txt` pair added per algorithm you claim
- [ ] Expected outputs are **verified by hand**, not copied from program output
- [ ] `mvn test -Dtest=GQLAlgorithmTest` passes
- [ ] The PR briefly explains why each expected result is correct on the test graph