A couple observations that have gone unnoted in the discussion:
1) Whether you like mirrors or not, when Canon opts to stop refreshing lenses in EF mount, it's game over. There is too much innovation going on in lens development (and a resulting quickening of the lowering of values of used equipment) for pros not to make the decision to move within 2-4 years after the inevitable trickling off. That point will, for Canon's business reasons, come sooner than many will wish, if it hasn't come already. We'll know only in retrospect. Almost certainly it'll be like Sony alpha mount, where no one will know if it is supported due to a dissonance between Canon words and actions, and then a few years later we'll realize it wasn't being supported for some time. Resale values of lenses will go down significantly, forcing pros concerned about capitalization of equipment to take the loss and move to mirrorless of some brand or other. I'm not advocating for this; just observing that I can't think of an exception to this mount evolution narrative in the past 30 years.
2) If you make more money on medical imaging than on cameras, your technology research emphasis will be on resolution, detection, optics, materials and a bunch of other things that really matter to those most important business lines. One of the areas not stressed is processor speed. Medical imaging is not a dynamic environment requiring tens of shots per second. Quite the opposite. You get a medical image, and they tell you to go home so some dude can come in and examine it and send you a report 2 weeks later. So if you want to control your own chip fab - a billions-dollars investment - you aren't necessarily multiplying the costs of that fab to maximize performance on an edge case product line like cameras. It may be the case that Canon is in a market position that requires a multiplication of computing needs just after a years-long period when the Digic chips haven't advanced nearly as quickly in computational capacity as those of other brands, perhaps because of a combination of build-it-here philosophy along with a corporate-wide prioritization of other research and capital needs. I do not know that this is the case, and I hope it is not. But I am aware and not overly-skeptical that it is a possibility. Were it true, you would see Canon releasing a series of cameras over a number of years that had lower frames per second than desired and versus the competition. This would be especially true on mirrorless. Where they did have great frame rates, you would see the use of multiple chips. You would also see computationally-intensive features, such as zebras, desirable codecs, etc. left out. I think you would also see an emphasis on the landscape and studio photographer for the "pro" mirrorless models, less the sports/action set. Were this incorrect, you would likely see evidence of that with one or two cameras that showed a physical and technological capacity beyond 300 mp/second throughput, like other brands have shown. Fingers crossed that's coming.