Add Vertex support

[em3s]

Summary

Add Vertex as a first-class entity type alongside Edge and MultiEdge. Vertex reuses the existing EdgeState encoding with source=key, target="-" (VERTEX_MARKER), so existing storage format and encoding pipeline are shared.

Key design decisions:

• LabelType.VERTEX added to codec-java so Vertex is identified by type.
• EdgeMutationStrategy.Vertex suppresses all EdgeCount writes (producesCount = false)

Closes #361

Changes

• codec-java: LabelType.VERTEX, BulkEdgeEncoder encodes State only (no index, no counter)
• core: ModelSchema.Vertex, TableDescriptor.Vertex, EdgeMutationStrategy.Vertex, VertexBulkMutationRequest, VertexMutationResponse, DataFrameVertexPayload
• engine: V3TableDescriptor.create() restores LabelType.VERTEX → V3TableDescriptor.Vertex on reload; V2BackedTableBinding routes Vertex to buildForVertex()
• server: VertexMutationController (POST /vertices, POST /vertices/sync), VertexQueryController (GET /vertices/get), schema/DDL wiring in TableCreateRequest, V3CompatService, V3MetadataConverter

How to Test

./gradlew test

New test coverage:

• VertexBulkMutationRequestTest: createEvent — source=id, target=VERTEX_MARKER, nullability validation, reserved id guard
• EdgeMutationStrategyTest / EdgeMutationBuilderTest: Vertex strategy — producesCount=false, no count records on CREATED/UPDATED/DELETED
• BulkEdgeEncoderTests: active/inactive Vertex → State only, no counter rows
• V3MetadataConverterTest: LabelType.VERTEX → TableDescriptor.Vertex round-trip
• MutationServiceSpec: Vertex insert/update/delete/get engine-level flow
• VertexIntegrationTest: full E2E — table create, mutation, get, update, delete

AI Assistance

• This PR was written largely with AI assistance.
  • Tool / model: Claude Code (claude-sonnet-4-6)

[em3s]

Usage Examples

0. Setup: Create database and table

# Create database
curl -X POST "http://localhost:8080/graph/v3/databases"   -H "Content-Type: application/json"   -d '{"database": "mydb", "comment": "my database"}'

# Create vertex table
curl -X POST "http://localhost:8080/graph/v3/databases/mydb/tables"   -H "Content-Type: application/json"   -d '{
    "table": "users",
    "schema": {
      "type": "VERTEX",
      "id": {"type": "string", "comment": "user id"},
      "properties": [
        {"name": "name", "type": "string", "comment": "display name"},
        {"name": "age",  "type": "long",   "comment": "age", "nullable": true}
      ]
    },
    "storage": "datastore://my_namespace/users",
    "mode": "SYNC",
    "comment": "user vertex table"
  }'

1. Insert vertices

curl -X POST "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices"   -H "Content-Type: application/json"   -d '{
    "mutations": [
      {"type": "INSERT", "vertex": {"version": 1, "id": "user1", "properties": {"name": "Alice", "age": 30}}},
      {"type": "INSERT", "vertex": {"version": 1, "id": "user2", "properties": {"name": "Bob"}}}
    ]
  }'

{
  "results": [
    {"id": "user1", "status": "CREATED", "count": 1},
    {"id": "user2", "status": "CREATED", "count": 1}
  ]
}

2. Update a vertex

curl -X POST "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices"   -H "Content-Type: application/json"   -d '{
    "mutations": [
      {"type": "UPDATE", "vertex": {"version": 2, "id": "user1", "properties": {"name": "Alice Smith"}}}
    ]
  }'

{"results": [{"id": "user1", "status": "UPDATED", "count": 1}]}

3. Delete a vertex

curl -X POST "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices"   -H "Content-Type: application/json"   -d '{
    "mutations": [
      {"type": "DELETE", "vertex": {"version": 3, "id": "user2", "properties": {}}}
    ]
  }'

{"results": [{"id": "user2", "status": "DELETED", "count": 1}]}

4. Insert (sync) — wait for write to complete

curl -X POST "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices/sync"   -H "Content-Type: application/json"   -d '{
    "mutations": [
      {"type": "INSERT", "vertex": {"version": 1, "id": "user3", "properties": {"name": "Carol", "age": 25}}}
    ]
  }'

5. Get vertices (multi-get)

curl "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices/get?id=user1,user2,user3"

{
  "vertices": [
    {"version": 2, "id": "user1", "properties": {"name": "Alice Smith", "age": 30}, "context": {}},
    {"version": 1, "id": "user3", "properties": {"name": "Carol", "age": 25}, "context": {}}
  ],
  "count": 2,
  "total": 2,
  "context": {}
}

user2 was deleted and is excluded. Unknown IDs are silently skipped.

[em3s]

Internal representation note

Vertex is stored as an Edge with source=id, target="-" (VERTEX_MARKER). The Edge API can reach the same row — this is not how you should use it, but it shows that the two types share the same storage format:

# Vertex API
curl "http://localhost:8080/graph/v3/databases/mydb/tables/users/vertices/get?id=user1"
# {"vertices": [{"version": 1, "id": "user1", "properties": {"name": "Alice"}, "context": {}}], ...}

# Edge API — same row, internal encoding exposed (do not use this way)
curl "http://localhost:8080/graph/v3/databases/mydb/tables/users/edges/get?source=user1&target=-"
# {"edges": [{"version": 1, "source": "user1", "target": "-", "properties": {"name": "Alice"}, "context": {}}], ...}

[em3s]

Verified locally — all curl examples in the comments above work as expected. Merging.