fix(devserver): use tablesSchemaV2 for Generate Data Model to include primary keys

[Funbucket]

Summary

Fixes #10517

When using Generate Data Model in Cube Playground, primary key columns are not marked with primary_key: true in the generated YAML, causing compile errors for cubes with joins defined.

Root Cause

DevServer.ts calls driver.tablesSchema() in both endpoints:

// /playground/db-schema
const tablesSchema = await driver.tablesSchema();

// /playground/generate-schema
const tablesSchema = req.body.tablesSchema || (await driver.tablesSchema());

tablesSchema() queries only information_schema.columns — it does not fetch primary key constraints. So no column gets attributes: ['primaryKey'], meaning ScaffoldingSchema never sets isPrimaryKey: true, and BaseSchemaFormatter never emits primary_key: true.

tablesSchemaV2() already exists and does the right thing: it calls primaryKeys() (implemented per-driver) and merges PK information into column attributes.

Fix

Replace both calls with tablesSchemaV2():

- const tablesSchema = await driver.tablesSchema();
+ const tablesSchema = await driver.tablesSchemaV2();

- const tablesSchema = req.body.tablesSchema || (await driver.tablesSchema());
+ const tablesSchema = req.body.tablesSchema || (await driver.tablesSchemaV2());

Impact

• Affects all databases where the PK column is not literally named id (the only fallback heuristic in ScaffoldingSchema)
• Reproduced with PostgreSQL 14 + AdventureWorks; any real-world schema with named PKs (salesorderid, productid, businessentityid, etc.) is affected

Test plan

• Generate Data Model on a PostgreSQL table with a named PK (e.g. salesorderid) — generated YAML should include primary_key: true
• Generated YAML compiles without "primary key is required" errors
• Existing behaviour for tables with no joins is unchanged