[FEATURE] Add json_extract function to PPL

[acarbonetto]

Is your feature request related to a problem?

As part of the RFC to add JSON functions, the json_extract function would be useful to extract specific objects and scalars from an existing json_object.

• META RFC: [META]Add PPL JSON extended functions support #3027
• PPL RFC: [RFC] Add PPL JSON extended functions support #3028
• Related Spark PPL PR: Support PPL JSON functions: construction and extraction opensearch-spark#780

What solution would you like?

### `JSON_EXTRACT`

**Description**

`json_extract(doc, path [, path]...)` Extracts a json value or scalar from a json document based on the path(s) specified.

**Argument type:** STRING, STRING

**Return type:** STRING | NULL | JSON ARRAY

- Returns a JSON array for multiple paths or if the path leads to an array.  
- Return null if path is not valid.

Example:

    os> source=people | eval `json_extract(json('{"a":"b"}'), '$.a')` = json_extract('{"a":"b"}', '$a')
    fetched rows / total rows = 1/1
    +----------------------------------+
    | json_extract('{"a":"b"}', 'a')   |
    +----------------------------------+
    | b                                |
    +----------------------------------+

    os> source=people | eval `json_extract(json('{"a":[{"b":1},{"b":2}]}'), '$.a[1].b')` = json_extract('{"a":[{"b":1},{"b":2}]}', '$.a[1].b')
    fetched rows / total rows = 1/1
    +-----------------------------------------------------------+
    | json_extract('{"a":[{"b":1.0},{"b":2.0}]}', '$.a[1].b')   |
    +-----------------------------------------------------------+
    | 2.0                                                       |
    +-----------------------------------------------------------+

    os> source=people | eval `json_extract('{"a":[{"b":1},{"b":2}]}', '$.a[*].b')` = json_extract(json('{"a":[{"b":1},{"b":2}]}'), '$.a[*].b')
    fetched rows / total rows = 1/1
    +-----------------------------------------------------------+
    | json_extract('{"a":[{"b":1.0},{"b":2.0}]}', '$.a[*].b')   |
    +-----------------------------------------------------------+
    | [1.0,2.0]                                                 |
    +-----------------------------------------------------------+

    os> source=people | eval `invalid_json` = json_extract(json('{"invalid": "json"'))
    fetched rows / total rows = 1/1
    +----------------+
    | invalid_json   |
    +----------------+
    | null           |
    +----------------+

What alternatives have you considered?

N/A

Do you have any additional context?

Related spark PPL PR: opensearch-project/opensearch-spark#780