but without income from customers interested in video, there's very little money for developing new sensors and processors!
not correct. There were 15 years of enormous progress in sensors and processors development used in DSLRs ... way before any of them was video-enabled. Ever since, most of the development money seems to be spent on this effort to try and turn still cams into half-assed videocams too.
All those who want to put sorry little video clips on youtube would be well enough served with their smartphones or compact digicams. And "serious" videographers at all levels from semi-pro to Hollywood-aspiring Indie filmmaker would be way better served with pure and true video cams. Sony, Pana, Canon could have started making large-sensor videocams a long time ago - customers were not holding them back.
Demand for HDSLRs was and is solely fueled by the fact, that they offered good video-quality for significantly less money than true videocams, but definitely not by overwhelming demand of users who wanted to have one combi-tool to capture video and stills with. It would be more than good enough, if both lines - stills and videocams - have the same or fully compatible lens mounts, so that quality
Video-enabled still cams are just being stuffed down photographers' throats ... if there was free choice between say a 5D-V (video+stills) and a 5D-S (stills only) for 20% less money I would expect that 80% of cameras sold would be the stills version. But again, we are beinmg denied that choice.
outdated, end of life DSLRs without video - e.g. 1Ds III - are no solution. But offer me a 1D-S (1D X minus video functionality) for 4k and I take it any day.
hammers for nails - screwdrivers for screws.
trucks for heavy loads - sportscars for speed.
videocams for video - still cams for still images!