[EPIC] Add semantic metadata management to Gravitino

### Describe the proposal

Gravitino provides unified management and governance for technical metadata, but it does not yet provide first-class management of business semantic definitions. Today, these definitions are fragmented across systems such as dbt, Cube, Spark/Databricks Metric Views, and Snowflake Semantic Views, making them difficult to discover, govern, and reuse consistently.

Add a vendor-neutral semantic metadata management layer to Gravitino. Semantic assets should support consistent identity, versioning, lifecycle, authorization, ownership, auditing, and discovery.

This epic focuses on semantic metadata management. Query compilation and execution, materialization, and engine-specific runtime integrations will be tracked separately.

### Task list

see below

Start by reading the proposal and its task list, which is currently empty, to understand the intended vendor-neutral semantic metadata layer. Done should cover consistent identity, versioning, lifecycle, authorization, ownership, auditing, and discovery while leaving query compilation, execution, materialization, and engine-specific runtime integrations separate.

Written by the indexing model from the issue text.