OK, that shouldn't matter, looking at it further, on my report it looks like it only updates the dependent parameters the first time, it may be related to this bug...

the main problem with using data from sys.dm_db_index_usage_stats directly is that the data there gets reset when the service restarts..  Do you know the uptime of this server?
[code...

how about something like this?
WITH invoice_sums AS
(SELECT invoice_number, passenger_name,
    SUM(fare) OVER (PARTITION BY invoice_number) AS total_fare,
    SUM(commission) OVER (PARTITION BY invoice_number) AS total_commission,
   ...

how do you have the StartDate and EndDate parameters setup?  I've seen something similar, where the StartDate and EndDate parameters have properties of:
Default Values: Specify Value, using a Switch...

I think the problem with linked servers isn't in how many you have setup, but the performance each time you use one.  I've seen systems where the performance limitation was...

OK, if CPU, memory, disks setup the same, have you checked the "cost threshold for parallelism" and "max degree of processing" on the 2 servers?

More importantly, looking deeper...

I notice on Server1 the execution plan went parallel in multiple places, does Server1 have more CPU cores in it than Server2?  That could be a contributing factor in it...

my favorite way to do this is to use CROSS APPLY to unpivot the data and then just use MAX():
DECLARE @yourtable TABLE
(colA INT,
colB INT,
colC...

Are you sure that tblB.id is the proper foreign key column to join bacl to tblA.id with?  If you join on the wrong column then yes the results for tblB...

Well I didn't really gain this level of knowledge just from books, it's more a mix of techniques from various resources and years of doing some trial and error experimentation...