I largely agree with using the right tools for the job. I also happen to be trained for industrial automation, so my perspective is a bit outside the normal for people regarding actually making stuff. One big issue is capital intensity. It's very expensive to buy an injection molder, and cutting tooling is also expensive. Unless you already do that, it's a massive amount of money to just buy molds and get a machine, and then a large investment in new tooling every so many parts or design iteration. I mean, you're looking at a large number of different parts, often ones with holes in multiple directions, meaning that you need additional cams. If you're a company that has injection molding equipment and tool-making gear, like SeemeCNC, given that Blackpoint engineering, which is where they came from, offered injection molding and tool-making as a service, it very rarely makes sense to print a production part (unless the chemistry is weird, or the geometry quite complex, such as the flexible parts for the Artemis).
I often see the average tooling cost at around twelve to fifteen thousand dollars. At retail price, just the tooling for a single mold would buy at least sixteen printers, and production, as well as improvement is more granular, although far slower and less efficient from a price perspective, as well as producing a more thermally cycled, and thus potentially less stable plastic.
The best solutions, as always, cost a LOT of money. the good, fast, or cheap mentality only gets you so far. In manufacturing, you have to balance part quality, production speed, reliability, lead time, cost, floor-space, labor, power requirements (Injection molders require serious amounts of compressed air and cooling water in the designs I've seen prints for), and ease of maintenance. Prusa picked something that gives them scalable production in a geometric fashion, as every 29 hours they could double the number of printers available to them, until they run out of machine tenders, or over-run the power to the building, or more likely, get mobbed for having not shipped any printers in quite some time. It is also quite possible that there simply was not a way to get an injection molding machine of any signifigant scale into their office building. Additionally, it would introduce more single points of failure (molds, molder, toolmaker, etc), which since they would likely not have the volume to justify technicians for those machines, with the need for only one or two of the things, might lay up production for a day or more should anything break.
Given say, three million dollars or a few mill more, I could buy a middle of the road injection molder, a machining center to make molds, as well as build a basic heat-treat oven to harden some of the molds to an OK, but honestly not stellar quality, as well as start doing reasonable labor reduction things, like having a bin under the molder that can be replaced while it's cycling, as well as a gantry crane to move molds between the machining center, heat-treat oven, and molder, as well as handling the bins of material. Then you need to hire tool-makers, operators, mold designers, and repair techs. For the size of niche Prusa is in, despite their quality, I'm not sure I would recommend they do IM in house, but they obviously want to do as much as humanly possible themselves, and I don't fault them for it.
Rostock Max V2, Duet .8.5, PT100 enabled E3D V6 and volcano, Raymond style enclosure
Automation Technology 60W laser cutter/engraver
1m X-carve router
Sic Transit Gloria Mundi