question

The event happens, as the name suggest, on model reset and then every 3600s/1h. Every time the event fires, the code in the Row Value(s) field is executed. You learn more about how row values work on the linked manual page in my original answer. In short: The function returns an array of values and for each element a new data row is created in the Statistics Collectors internal table.

The code reads the Input stat from the objects and stores it in a map label, so when the event fires the next time, the previous Input value can be read from there to calculate the difference the previous and current value - the input that happened in the meantime.

Both a reference to the object and the calculated difference are written to the row value and then used in the Columns tab to write this data to the table.

Below is a commented version of the code.

// Get an array of all objects from the group linked through the "Objects" label
Array storages = collector.Objects.as(Group).toFlatArray();
// Create row values in format [storage, last hour input]
Array rowValues = [];
// For each object in the array
for(int i = 1; i <= storages.length; i++)
{
    Object storage = storages[i];
    // Read the previously measured input value that is stored in the "QtyMap" label for this object
    int prevInput = collector.QtyMap.as(Map)[storage];
    // Get the current input statistic
    int curInput = getstat(storage, "Input", STAT_CURRENT);
    // Update the QtyMap label to the current value
    collector.QtyMap.as(Map)[storage] = curInput;
   
    // Push a reference to the object and the difference in input between last time and now to the row value array
    rowValues.push([storage, curInput - prevInput]);
}

// Each entry in the row value will create a new row (Each entry is an array with two element)
// These are accessible through "data.rowValue" when writing data to the column
return rowValues;

You could also have the table view and labels tab open while the model is running to see how they change when the event fires.

Note how the values for 10:00:00 are the difference between the previous values and what you see in the label map (which holds the Input stat value from the latest measurement, not necessarily in order).