How to unnest a json arrays columns

I have a column called event_params with values like this:

[{"key": "label", "value": {"int_value": null, "float_value": null, "double_value": null, "string_value": "example"}}, {"key": "label2", "value": {"int_value": null, "float_value": null, "double_value": null, "string_value": "another_example"}}, {"key": "label3", "value": {"int_value": 9, "float_value": null, "double_value": null, "string_value": null}}, {"key": "label3", "value": {"int_value": 1234, "float_value": null, "double_value": null, "string_value": null}}]

OR

[{"key": "label",
  "value": {"int_value": null,
            "float_value": null,
            "double_value": null,
            "string_value": "example"}},
  {"key": "label2",
   "value": {"int_value": null,
            "float_value": null,
            "double_value": null,
            "string_value": "another_example"}},
  {"key": "label4",
   "value": {"int_value": 9,
            "float_value": null,
            "double_value": null,
            "string_value": null}},
  {"key": "label4",
   "value": {"int_value": 1234,
             "float_value": null,
             "double_value": null,
             "string_value": null}}]

I would like to unnest this column to extract the string_value from key = "label" for every user_id. To do that I'm trying to use jsonb_to_recordset function but the output I'm getting is basically the same user_id but multiple times and the only thing it changes it's the values from each "value" key:

user_id | value
123 | {"int_value": null, "float_value": null, "double_value": null, "string_value": "example"}
123 | {"int_value": null, "float_value": null, "double_value": null, "string_value": "another_example"}
123 | {"int_value": 9, "float_value": null, "double_value": null, "string_value": null}
...

All of that for the same user even though there are more users (I would expect to have just one row per user). Here is the piece of code that I'm using:

select
	user_id,
	event_params,
	evs.key,
	evs.value::json
from google_analytics.firebase_events_analytics,
	jsonb_to_recordset(event_params) as evs(key text, value text)

The query is returning multiple rows per user_id because of an implicit CROSS JOIN. The following is the equivalent when the CROSS JOIN is explicitly expressed (the implicit LATERAL associated with JSONB_TO_RECORDSET is also included):

SELECT user_id,
       event_params,
       evs.key,
       evs.value
  FROM google_analytics.firebase_events_analytics
  CROSS JOIN LATERAL JSONB_TO_RECORDSET(event_params) AS evs(key text, value jsonb)

The desired behavior can be achieved by filtering the joined rows. The following uses criteria in the WHERE clause to exclude undestired rows:

SELECT user_id,
       event_params,
       evs.key,
       evs.value
  FROM google_analytics.firebase_events_analytics
  CROSS JOIN JSONB_TO_RECORDSET(event_params) AS evs(key text, value jsonb)
  WHERE evs.key = 'label';

An alternative is to use JOIN criteria:

SELECT user_id,
       event_params,
       evs.key,
       evs.value
  FROM google_analytics.firebase_events_analytics
  JOIN JSONB_TO_RECORDSET(event_params) AS evs(key text, value jsonb)
    ON evs.key = 'label';

The two approaches are equivalent and produced identical explain plans when I tested the queries.