Once you get within ~.002-.003 all over on a large area (if you can get that close) on a turret mill, that is about as good as it gets unless the machine is brand new and everything is "perfect". That barely covers the tooth mark depth for a lot of situations. Your part looks that close, .003" anyway. From what I can see from my computor screen anyway
Even if your part has stress in it, that is not the problem today if you have already made several passes and still get the same results. It will move the first roughing pass, a little more the next, but by the time you are merely skimming .002 at a time, it is not moving "right now" though it might again tomorrow. But the point is, when it moves, the marks won't be the same you "keep" getting off the machine.
What machine are you milling it on? It is somewhat unusual for a turret mill to make a casting hollow. The way they wear, usually they face slightly convex.
As Richard notes, if you are going to invest the time to make this a precision tool (meaning you are going to have to scrape it) then by all means rough machine it and then have some form of qualified stress relief done. I used to make and sell SE's and I would get them from the foundry, rough machine them "all over" and take them to Elmira Heat Treat for thermal stress relief. At that point, they usually came back about .005-.007 out of flat ~3" x 30" long. At that point, I sold them because you can imagine doing a further light machine and another stress relief could add 50% or more to the cost. Also, most people want different angle configurations, etc. so it was pointless to charge everyone for something only a few would use. For myself, and I have a planer, it was seldom worth it to even mount the casting back on a machine at that point. Knocking even .007" off the rocking point(s) as a rough scraping operation and getting down to ..002 (where it would reliably come off the machine without significant time and effort) went faster than the machine set up and time, and puts less stress back in the casting.
Every time you machine a casting, it puts some stress in it.
All these reasons are why precision surfaces and alignments are scraped. If you have a really good grinder and perfect process control, you can grind. But that puts stress in, too. And the average job shop grinder is still going to yield a surface that will blue up about like you show, though the low and high spots will only be a few or single tenths different. This will be a combination of thermal issues, stress in the casting issues, machine imperfections, foundation on which the machine sets, and operator control/setup/inputs including how it is mounted to the magnet and residual magnetic influence. That blue will show 50 or even 10 millionths, if you are using it thinly.
You can probably have that part hitting all over in less than 10 minutes with a scraper, except there will be some holes, and the bearing quality will be coarse.
Point being, the reason people scrape at a given point is not because they love doing it and like to give so much money to the chiropractors, but because at a certain point scraping is just way faster and more efficient that trying to machine any closer. If you have a way grinder and maintain it every day and keep good process control, you can get within tenths on really big areas and just flake for bearing/oil/knock off the few obvious contact points and everything else is flat. In the small shop, you are probably going to have to scrape when you get to within .001 on a small casting, and .003-.005 or so on something "large" off a turret mill.
You are closer than that, though it will probably take .003 - .005 on that to get flatness and get under all the residual tool marks. That is very quick work with a biax and probably still faster even by hand than to keep trying for "better" given your machine, existing set ups, and experience so far.
smt