I did happen to see the subquery recommendation, but that still didn't work. I checked the execution trace and you can see the subquery implemented. Its running on the datasource SQL server:
```
SELECT s0.searchparameter, s0.oid_site, s0.equipment,
s0.equipmentserialnumber, s0.activity, s0.duration, s0.durationunit,
s0.endtimedatetime, s0.starttimedatetime, s0.cycleidentifier,
s0.channel_name, s0.iteration_time, s0.p10, s0.p20, s0.p30, s0.p40,
s0.p50, s0.p60, s0.p70, s0.p80, s0.p90, s0.topsize, s0.s_36, s0.s_24,
s0.s_18, s0.s_15, s0.s_12, s0.s_10, s0.s_9, s0.s_8, s0.s_7, s0.s_6,
s0.s_5, s0.s_4, s0.s_3, s0.s_2, s0.s_1, s0.s_050, s0.s_025,
s0.aspect_ratio_sample_average, s0.aspect_ratio_std_deviation,
s0.sd_ratio_10_5, s0.image_rank FROM (SELECT t0.searchparameter AS
searchparameter, t0.oid_site AS oid_site, t0.equipment AS equipment,
t0.equipmentserialnumber AS equipmentserialnumber, t0.activity AS
activity, t0.duration AS duration, t0.durationunit AS durationunit,
t0.endtimedatetime AS endtimedatetime, t0.starttimedatetime AS
starttimedatetime, t0.cycleidentifier AS cycleidentifier, t0.channel_name
AS channel_name, t0.iteration_time AS iteration_time, t0.p10 AS p10,
t0.p20 AS p20, t0.p30 AS p30, t0.p40 AS p40, t0.p50 AS p50, t0.p60 AS p60,
t0.p70 AS p70, t0.p80 AS p80, t0.p90 AS p90, t0.topsize AS topsize,
t0.s_36 AS s_36, t0.s_24 AS s_24, t0.s_18 AS s_18, t0.s_15 AS s_15,
t0.s_12 AS s_12, t0.s_10 AS s_10, t0.s_9 AS s_9, t0.s_8 AS s_8, t0.s_7 AS
s_7, t0.s_6 AS s_6, t0.s_5 AS s_5, t0.s_4 AS s_4, t0.s_3 AS s_3, t0.s_2 AS
s_2, t0.s_1 AS s_1, t0.s_050 AS s_050, t0.s_025 AS s_025,
t0.aspect_ratio_sample_average AS aspect_ratio_sample_average,
t0.aspect_ratio_std_deviation AS aspect_ratio_std_deviation,
t0.sd_ratio_10_5 AS sd_ratio_10_5, dense_rank() OVER ( PARTITION BY
t0.cycleidentifier ORDER BY t0.iteration_time ASC ) AS image_rank FROM
T_175690704883147211741866734301961774174722148934560550270244 t0) s0
WHERE s0.searchparameter = ? AND s0.oid_site = ? AND s0.image_rank = ?
```