I am interested also in this topic. Recently I was making some tests with the PRO-1000. I believe that in the base media types are coded some kind of recipes of ink mixing. I have two examples to support that.
I was trying to replicate "Pro Luster" media setting. I ended with "Lightweigth Photo Paper A" as the closest contender. I set everything as the "Pro Luster". When I used a icc profile made with "Pro Luster" media setting the colors went way off. The blue sky went magenta.
Second I was interested after reading Mark Segal review of Canon Fine Art Smooth paper to try it. I haven't managed yet to find it here but I started to play with the "Fine Art Smooth" media setting. I check it's settings and were the same as the "Highest Density Fine Art Paper". But if the settings are the same then what is the need of a specific media type? So I printed the same B/W image with the printer B/W mode with the same settings on the same paper but with the media setting first "Highest Density Fine Art Paper" and then with the "Fine Art Smooth". The difference was easily visible with the blacks darker with the "Fine Art Smooth". So I checked the ink consumption for the two prints and I found that with the "Fine Art Smooth" media setting the printer decreases the use of the two grays and increases all other colors. See the details in the attached screenshot.
Yes, it's interesting what lays behind these differences in media types. You might assume different media types to behave similar, with all settings reported by MCT being the same, and printing with identical driver settings and icc - but not the case as you mention.
Not very transparent, so hard to know how and what the choice of the OEM or user made media types affects.
Also curios to note the amount of PBK used in your two print examples at the end. Both used media types are set to MBK, not the 'MBK,PBK' mix, but apparently it mixes anyhow. So very hard to know what's going on, or indicated in MCT (Media Configuration Tool).
BTW, MCT has several bugs still. One is when doing calibration - it prints a feed adjustment prior to the calibration, even though that's already done separately in step 1. And what's worse, it gives a different result than the manually initiated one - so which one counts. Can anyone read out a media type amx file and find the used value? Then I can see which of the two feed adjustments are actually recorded.
Also I just noticed a custom media type with unique calibration done has forgotten it's calibration. Now just says it's using the common calibration from a Pro Luster calibration.