AggregatingMergeTree order by column not in the sorting key - clickhouse

Question

What are some options to have AggregatingMergeTree merge by a column but ordered by a column that's not in the sorting key?
My application is similar to Zendesk tickets. A ticket has a category, status, and ID. The application emits ticket status change events to CH and I'm calculating statistics on the time it took to close since it was created given some time range R group by some time period P.
For example, events look like this
{
"ticket": "A",
"event_time": 2022-12-08T15:00:00Z,
"category": "bug",
"status": "created"
},
{
"ticket": "A",
"event_time": 2022-12-08T15:30:00Z,
"category": "bug",
"status": "reviewing"
},
{
"ticket": "A",
"event_time": 2022-12-08T16:00:00Z,
"category": "bug",
"status": "reviewed"
}
My AggregatingMergeTree (more specifically, it's replicated) has a sorting key on the ticket ID to aggregate two states into one.
CREATE TABLE ticket_created_to_reviewed
(
`ticket` String,
`created_ticket_event_id` SimpleAggregateFunction(max, String),
`created_ticket_event_time` SimpleAggregateFunction(max, DateTime64(9)),
`created_ticket_category` SimpleAggregateFunction(max, String),
`close_ticket_event_id` SimpleAggregateFunction(max, String),
`close_ticket_event_time` SimpleAggregateFunction(max, DateTime64(9)),
`close_ticket_category` SimpleAggregateFunction(max, String),
)
ENGINE = ReplicatedAggregatingMergeTree('<path>', '{replica}')
PARTITION BY toYYYYMM(close_ticket_event_time)
PRIMARY KEY ticket
ORDER BY ticket
TTL date_trunc('second', if(close_ticket_event_time > created_ticket_event_time,
close_ticket_event_time, created_ticket_event_time)) + toIntervalMonth(12)
SETTINGS index_granularity = 8192
Two MVs SELECT on the raw events and inserts to the ticket_created_to_reviewed. One for WHERE status = 'created' and another for WHERE status = 'reviewed'
So far the data populates correctly, although I have to exclude rows that only have one of the status events populated. Getting hourly p99 of ticket time to close past 1 day for each category looks something like this
SELECT
quantile(0.9)(date_diff('second', created_ticket_event_time, close_ticket_event_time)),
date_trunc('hour', close_ticket_event_time) as t,
close_ticket_category as category
FROM
(
SELECT
ticket,
max(created_ticket_event_id) AS created_ticket_event_id,
max(created_ticket_event_time) AS created_ticket_event_time,
max(created_ticket_category) AS created_ticket_category,
max(close_ticket_event_id) AS close_ticket_event_id,
max(close_ticket_event_time) AS close_ticket_event_time,
max(close_ticket_category) AS close_ticket_category
FROM ticket_unreviewed_to_reviewed
GROUP BY ticket
)
WHERE close_ticket_event_id != '' AND created_ticket_event_id != '' AND
close_ticket_event_time > addDays(now(), -1)
GROUP BY t, category
The problem is close_ticket_event_time is not in the sorting key so the query scans the full table, but I can't also include that column in the sorting key because the table wouldn't then aggregate by the ticket ID.
Any suggestions?
Things tried:
Adding an index and/or projection that orders by close_ticket_event_time. However, I think the main problem is that the sorting key is on ticket ID so the data is not ordered by time to efficiently find the matching time range, but at the same time adding close_ticket_event_time breaks the aggregation behavior in AggregatingMergeTree
MV that joins created ticket and closed ticket, and a different destination table with close_ticket_event_time as the sorting key. The destination table doesn't contain all the data if the right side of the JOIN isn't available at the time MV was triggered (i.e. left side). This can happen if events are ingested out of order.
Ideally, what I'm looking for is something like this in AggregatingMergeTree, but it appears this isn't possible due to the nature of how the data is stored.
PRIMARY KEY ticket
ORDER BY close_ticket_event_time
Thanks in advance