cbDrift

I would expect if you just set a condition condition to something that will always be true such as "MAP >1kpa", it will start as soon as you connect.  

Knock level global and knock level detected are not very useful runtimes when using the (older) ign/fuel trim strategy as the knock level global is basically the raw signal from the sensor with no windowing applied as far as I remember.  I cant remember if it has the frequency filter applied or not either.  
Knock level detected was originally added for the normalised mode which uses knock level normalised so is more useful with that strategy. 
"Knock count global" will increment if any individual knock level exceeds the threshold so that is a good one to monitor for a quick overview where/if knock has occurred.  
 

This is how normalised mode works. 
With normalised mode the ecu is comparing the noise level of the current combustion event to the previous few "learnt" combustion events.  If you are operating under reasonably stable conditions then suddenly increase load or RPM, then the first few combustion events at that new higher load will now be generating more noise than the previous events it is comparing to.  So to account for this, when there is a sudden change in load or RPM (this is determined by the RPM/MAP/TPS delta level settings, then the threshold is increased by the "Threshold gain" for the number of cycles set in "Delta hold".   
With normalised mode you would usually have a "flat" threshold table, and the threshold should only need to step up for a short time when there is a significant change in load or RPM.  You should not need to increase threshold with RPM or boost like your threshold table is, this suggests something is not right in the setup or possibly mechanical.  I notice you have a 2.5L whereas our car was a 2.0 so possibly the knock frequency is not ideal.  You could try it at 6,7 and say 13Khz to see if the noise levels look less erratic.  Your ignition advance is quite low so I doubt there is knock, but it is a little odd that your cyl 4 knock level is mostly the one that is erratic.  
A quirk I found in our subaru was the sharp click from the AC clutch turning on/off in auto mode would cause a spike in knock level, I got around this by disabling knock for a few tenths of a second before the clutch switched on of off.  I see that has caused 1 spike in your log but it is not your main issue, just something to keep in mind if you start seeing random spikes or retard.  

we use some eastern european standalone controllers for the dccd. in which you can alter settings on tps, 4 wheel speeds , steering angle etc. everything comes over canbus i believe. so to my understanding there's not gonna be much difference if you change only the engine ecu which emulates the standard canbus.

You should read the help page; PCLink G5 User Manual > User Interface > Layouts and Views > Views to get a more complete answer, but as a quick comment, if you want more than one settings view then you need to lock the existing one (click the padlock at the top left corner of the view).  Further, if you want 2 copies of the same table (for example one in tabular view and one in surface view) then lock the first one and shift click to add the second. 
You can also hold down ctrl when clicking a ecu settings tree menu to get the whole menu in one settings view.  

Our test car was a 2.0JDM GRB, dead stock including air box, MAF, exhaust, injectors and top mount IC.  
I never got around to removing the injectors to characterise properly so there could be some deadtime error baked into the tune which would consequently bake some error into the test you do for the fuel charge cooling coeff, so I cant say for sure that is representative.  Be aware injector spray pattern affects charge cooling a lot also.  having said that, the fuel control was quite stable in that car, the long-term fuel trims would average around +/-3% whenever I looked at them over different seasons and varied driving conditions.   I mostly ran the Mobil 98 in that car.  
I use the test described in the help file for the charge cooling coeff to approximate both the deadtimes and the charge cooling.  Ideally you need to do this on the dyno while holding steady and light load to get the PW up around 2ms where things are linear, I then set lambda target to 1.0, set the fuel table so it is flat around the operating area and lambda is sitting close to 1.0.  Then change the lambda target to 0.95, note how much measured lambda changes, then change target to 1.05 and again note measured lambda.  If it overshoots/undershoots the target significantly more on one side of 1.0 than the other, then the charge cooling needs adjustment.  Tweak that and try again until you get a similar amount of over/under correction when you step the target leaner or richer from 1.0.  Then do the same but tweak the deatime with each test until the measured lambda no longer over/undershoots a change in target.  This sounds complicated but its really only about 15mins work once you get the hang of it and have everything set up to display and adjust all on the one page.
The charge temp approximation table worked well in that car with the top mount, I dont think a front mount would change it significantly unless the manifold gets significantly cooler with the top mount removed.  We were using the stock JDM sti IAT location beside the throttle.