Personally, I think this is actually good. Consistent and Productive Design, especially on the level of patterns across an entire operating system of both first and third party applications, is an insanely difficult problem. This means one absolutely needs to bring in the consideration of resource allocation. Its not possible to do everything exactly the same; in fact, a related but incorrect argument may be that its not possible to do everything exactly the same on Day 1, but because "Good Design" is a moving target, its actually really difficult to reach that goal in any capacity whatsoever.
Inconsistent design, like this, tells me that it isn't a resource allocation priority for Microsoft. Thus, those resources are most likely diverted to more functional aspects; getting it working, faster, more consistent behavior, broader and deeper feature sets, etc.
This is the right trade-off for Windows. One can point to MacOS or iOS and see a much more consistent design language across first-party apps, as well as pretty strong adoption of this design language across third-party apps, but the cost of this near-perfection is abstracted away. Its easy to see inconsistencies in design; its much harder to speak in aggregates about obscure software issues, missing features, delayed releases, etc. Strongly consistent design, in the face of these issues (many of which are unproveable externally) may actually be the right trade-off for MacOS and iOS; Apple is a far more design/PR/marketing driven company, and this guise of "everything is engineered and under control" may actually be materially important for their brand.
Windows is different. Moreover; it should be different. The fundamental idea that an entire operating system needs a cohesive, strongly consistent design language is one that Microsoft has tried to push several times, but one which hasn't even seen internal adoption, let alone external. One can blame this on the design languages being bad, or their tooling being bad, or anything, but at the end of the day its still the most popular non-server OS in the world (or is Android bigger now? Well, its the same argument there too). Conforming to any central design pattern isn't a priority for Windows users. There's no physical law of the universe that OS-wide design languages are a naturally good thing; in fact, I would argue that design languages shared between many domains of applications may fundamentally fail to account for the wide variety of interaction patterns within every domain, and thus as a software platform increases its complexity of services offered, domain-agnostic design languages naturally begin to fall apart. They may actually be Generally Bad, though I only feel comfortable asserting with any certainty that they're probably not Generally Good or Bad, and there's room for different platforms to try different things.