I notice that the version you uploaded was set to run in 2022.1.2
Here are some data files running in 2024.2.4. One using your code, one setting start times to frame 1 instead of 0 seconds and one with a blank text component set to start at frame 1 with a duration of 1 frame before each of your routines.
There seems to be an overhead for starting a loop so one option for you would be to present your rows using a randomised list which places T1 items at the correct intervals. I use a similar method in my Complex Randomisation online demo PsychoPy Online Demos
1 frame blank_T1_2024-11-05_13h24.38.533.csv (16.1 KB)
2024.2.4_T1_2024-11-05_13h13.09.592.csv (13.5 KB)
by frame_T1_2024-11-05_13h18.54.459.csv (13.5 KB)
024033_T1_2024-11-05_13h11.12.637.csv (9.6 KB)