Define how JSON database types should be mapped to DataFrame (String vs ColumnGroup)

### Problem

When reading data from JDBC, JSON-typed columns are currently represented as `String`.

However, JSON values naturally map to hierarchical data structures, and representing them as plain strings loses structure and makes further processing harder.

### Design question

How should JSON database types be represented in DataFrame?

Options:

- keep current behavior (JSON as `String`)
- automatically parse JSON into `ColumnGroup`
- provide an opt-in parsing mode (e.g. flag or configuration)
- provide helper API for explicit conversion

### Expected

Define a consistent mapping strategy for JSON database types.

### Acceptance criteria

- Decision made on default behavior (String vs structured)
- If structured mapping is supported:
- JSON is parsed into `ColumnGroup`
- nested structure is preserved
- If not:
- provide explicit API or document recommended conversion
- Behavior is documented with example

### Motivation

DataFrame supports hierarchical data structures such as JSON

Returning JSON as plain strings prevents users from using the full capabilities of the API and requires manual parsing.

The existing solution supports only presenting JSON type as a String
We need to convert correct JSON strings to the hierarchical structure after reading from the Result Set