feat: support vector indexes via ALTER TABLE ... CREATE INDEX

[wombatu-kun]

This is one of three PRs that supersede #436, splitting it per reviewer feedback.

Summary

Extends the existing ALTER TABLE … CREATE INDEX … USING method (...) statement (grammar rule LanceSqlExtensions.g4#createIndex) to accept vector index methods (ivf_flat, ivf_pq, ivf_hnsw_pq, ivf_hnsw_sq) alongside the existing scalar methods (btree, fts). No new SQL statement is introduced — the grammar rule is unchanged; only the method parameter accepts new values, exactly as the reviewer suggested.

Example:

```sql
ALTER TABLE lance.db.items CREATE INDEX emb_idx USING ivf_pq (embedding)
WITH (num_partitions = 256, num_sub_vectors = 16, metric = 'cosine');
```

Vector index training currently runs single-shot on the driver (`AddIndexExec.runVectorIndex`) because Lance's distributed vector-index path requires pre-computed IVF centroids — per-fragment tasks cannot train a global codebook on their own and the native code rejects the call with "Build Distributed Vector Index: missing precomputed IVF centroids". A follow-up can precompute centroids in a Spark job and re-enable the per-fragment build via `IvfBuildParams.Builder.setCentroids`.

`DistanceTypes` is shared infrastructure for parsing user-facing metric strings (`l2` / `cosine` / `dot` / `hamming`) into the `DistanceType` enum from lance-core.

Index correctness is verified through the existing DataFrame API path (the `"nearest"` read option backed by `LanceSparkReadOptions.CONFIG_NEAREST`), so this PR has no dependency on the SQL TVF PR — the two can be merged in any order.

Test plan

• `make test SPARK_VERSION=3.5 SCALA_VERSION=2.12 -Dtest=LanceVectorIndexTest` — 6 integration tests pass (1 skipped on 3.x: float16 needs Arrow 18+)
• `make test SPARK_VERSION=3.5 -Dtest=VectorIndexParamsBuilderTest` — 9 unit tests pass
• `make lint` — checkstyle + spotless clean
• `make test-all` — please run on CI to verify all version modules

🤖 Generated with Claude Code