Removing blinks affects multiple Raw files

Hello BSt support:
I am relearning how to set up a processing pipeline. I was able to create a sequence as follows: baseline removal, bandpass filter, notch filter, and resample. I saved this pipeline, then ran it on a single subject's raw EEG file. It created 4 new nodes, each with a further processing step in the node name. I looked at the EEG from each of these 4 nodes and they looked correct.
Then I cleared the Process 1 box and dragged the last node to it. My intent was to operate only on this last node. I read up on the blink removal, and manually ran the two steps needed from the Event box artifact tab: detect eye blinks and SSP EOG blink. Then I looked at all 5 nodes and ALL 5 of them had had their blinks removed. This does not seem right to me.
Here are some screen shots:
after running saved pipeline (bandpass, notch, resample)

some steps:


after blink removal:

I also went back and checked the first step of the pipeline, DC removal, and it had not been performed. I thought it had but I was using the Dc remove display option the first time I looked at it.
Maybe I do not fully understand the link to file aspect of all this?
Finally, I was confused by the Projector component screen - where is this documented?

This happened as when the Subject was created, the option ** Yes, use one channel file per subject**. This option is not recommended and prompts this warning:

As you can see, the linear operations (aka projectors) are stored in the channel file. Thus in your dataset, although the projector is computed for the pre-processed recordings, the projector is saved in the common channel file, thus available for all the other recordings using the same channel file.

This may lead to some confusion and sometimes to manipulation errors. For this reason, we decided not to recommend this setting.
From this described in the Create a new protocol tutorial:

This documented in the ** Artifact cleaning with SSP**, keep in mind that this tutorial relies on concepts of the previous tutorials

Thank you. I was focused on the desire to use a single generic electrode and head model for all subjects and all runs since I will not be able to measure either. I glossed over the specific warnings as I was not thinking I would be using SSP or other operators.