Am I missing something, what exactly are these "25 ip filter rules" and "25 simple queues" that cause such dramatic changes in Test Results of all Mikrotik routers?

I can find no further information about them, yet at the same time we seem to be encouraged to perform similar performance testing using the Traffic Generator.

I would also like to better understand these performance tests. (Somewhere I read about the methodology of the test but I do not remember where).
Something I noticed is that they do not involve access via Wireless for example. (how do you test on devices where you only have an ethernet interface?)
Many people complain that uses BTEST and performance is not even near to what is disclosed.

They are, exactly, this: 25 ip rules (firewall rules) and 25 simple queues. The idea is to give some measure of performance with load. So, You have the results with no rules, no rules with fast path (best case possible), with 25 firewall rules and with 25 simple queues.

The methodology is stated under the table. It even says the relevant RFC.

I read the RFC, and it does give a clearer indication of the general process. But it says nothing about queues and also states:

An IP rule is just it: a rule inserted in the firewall, like one allowing ssh from a given IP. Each firewall rule taxes the CPU. Insert enough rules, and even a CCR1072 won't be able to keep up with a slow network. They did one of the tests with 25 such rules - this way You can compare the results with and without them.

A queue is used to shape traffic. Take a look at the manual: there You will find all about them. Simple queues are easier on the CPU - and quite common. They do the same: show the speed of the router with 25 such queues.

The idea is the same, on all cases: to give You a handle on the capabilities of the hardware. It is impossible to show an example to each possible configuration, so they did it this way.