How to make JSON-like result merging ROW() structs into ARRAYs using GROUP BY ?

Suppose I have:
SELECT * FROM t ORDER BY id, tag;
id | tag | value
----+-----+-------
1 | a | 0.1
1 | b | 0.2
2 | a | 0.15
2 | b | 0.25
2 | c | 0.35
2 | d | 0.45
(6 rows)

I want to generate something JSON-like/compatible:

id result
---+-------------------------------------
1 | [{"a":0.1},{"b":0.2}]
2 | [{"a":0.15},{"b":0.25},{"c":0.35},{"d":0.45}]

I can achieve some things that are close:

WITH r AS (
SELECT id, tag || ':' || VALUE::VARCHAR kvp FROM t
)
SELECT id, IMPLODE(kvp) WITHIN GROUP (ORDER BY id, kvp)
FROM r GROUP BY 1 ORDER BY 1;
id | IMPLODE
----+---------------------------------------
1 | ["a:0.1","b:0.2"]
2 | ["a:0.15","b:0.25","c:0.35","d:0.45"]
(2 rows)

SELECT id, STRING_TO_ARRAY(
LISTAGG(tag || '|' || CAST(value AS VARCHAR) USING PARAMETERS separator='|') WITHIN GROUP (ORDER BY tag)
USING PARAMETERS collection_delimiter = '|')
FROM t
GROUP BY 1 ORDER BY 1;
id | STRING_TO_ARRAY
----+-----------------------------------------------
1 | ["a","0.1","b","0.2"]
2 | ["a","0.15","b","0.25","c","0.35","d","0.45"]
(2 rows)

But I can't find a way to create structs, then merge them into an ARRAY.

SELECT id, (tag, value)::ROW(VARCHAR,VARCHAR) FROM t ORDER BY id, tag;
id | row
----+----------------------------
1 | {"tag":"a","value":"0.1"}
1 | {"tag":"b","value":"0.2"}
2 | {"tag":"a","value":"0.15"}
2 | {"tag":"b","value":"0.25"}
2 | {"tag":"c","value":"0.35"}
2 | {"tag":"d","value":"0.45"}
(6 rows)

SELECT id, LISTAGG((tag, value)::ROW(VARCHAR,VARCHAR)) FROM t GROUP BY id ORDER BY id, tag;
ERROR 3457: Function LISTAGG(record) does not exist, or permission is denied for LISTAGG(record)
HINT: No function matches the given name and argument types. You may need to add explicit type casts

SELECT id, LISTAGG(CAST((tag, value)::ROW(VARCHAR,VARCHAR) AS VARCHAR))
FROM t GROUP BY id ORDER BY id;
ERROR 2366: Cannot cast type ROW(varchar(80),varchar(80)) to varchar

SELECT id, IMPLODE((tag, value)::ROW(VARCHAR,VARCHAR))
FROM t GROUP BY id ORDER BY id;
ERROR 9824: Implode is not supported for arguments of struct types

Perhaps this can't be done, since if you use AS to name fields in an array of ROW elements,
and the fields differ across the elements, Vertica forces the right-most name on you.

SELECT ARRAY[ROW('.1' AS a), ROW('.2' AS b)]
UNION ALL
SELECT ARRAY[ROW('.15' AS a), ROW('.25' AS b), ROW('.35' AS c), ROW('.45' AS d)];

array

[{"d":".1"},{"d":".2"}]
[{"d":".15"},{"d":".25"},{"d":".35"},{"d":".45"}]
(2 rows)

Seems like it should be possible.