I have a dataset as below

[
env_seqNum:int, env_tim:datetime, message:string, scenario: string, subscenario: string, requestId: string
1,2024-05-30T09:33:29.00Z, StartingOperation getToken XXX, ScenarioA, SubscenarioA, abcd-fghf,
1,2024-05-30T09:33:35.00Z, FinishingOperation getToken XXX, ScenarioA, SubscenarioA, abcd-fghf,
1,2024-05-30T09:33:55.00Z, StartingOperation getToken XXX, ScenarioA, SubscenarioA, abcd-fghf,
1,2024-05-30T09:33:58.00Z, FinishingOperation getToken XXX, ScenarioA, SubscenarioA, abcd-fghf,
]

As you can see, I have a token acquire operation thats done twice in same request for doing two API calls.
Now, I want to know the cumulative time thats taken for this token acquistion operation per request.
I cannot just do a diff of first and last row as it gives me a diff of 29 secs while in reality both token acquire operations took 9s (6 & 3)
How do I get this working in KQL?