Blender developer doesn’t owe you anything. Take it or leave it.
While I do understand some of the technical and budget limitations, I do feel it was a mistake to kill internal. You can never have enough render options. Burning that bridge removed a lot of features for quite a while. At my studio we had to use legacy versions of Blender along side the new one for a while. Even now I am considering comping halo into a scene from internal render.
Would be very useful to reach back into that bag of tricks and not have to use volumes which are noisy and expensive. And then just comp right in Blender.
I agree, internal was technically viewport too. Modern workbench is a lot slower in comparison to internal. It ofc has cool features but when it comes to frames per second internal do to its simplicity was way ahead.
Workbench is supposed to become faster this year, after Vulkan support.
We are still waiting for GPU acceleration for OpenSubdiv.
Abandon was the only choice possible if we take into account to their manpower.
Maintaining BI would have as consequence a slower EEVEE/Cycles development.
Original plan was to provide a Viewport Compositor.
I suppose that Halos would have been an effect supported by it.
Halos are a post-processing effect for BI.
It probably could be restored as a post-processing effect for other engines in 2.8 without reintroducing the whole BI.