Another thing on this topic of valve seal. Once I had some tester valves to act as my standard, I could safely assume the cages were sealing properly. But I was getting inconsistent results on the valves, so I knew there was room for improvement there. No matter how the 45-deg valve face was made on the lathe, I could spot 'record player' tool marks under magnification. The surface looks shiny & true to the naked eye but these undulations are enough to cause seal issues.
So my method was quite simple. For finishing I decided to get the got the valve off the lathe and into a cordless power tool (like a Dremel) on low speed. Blue the face & basically take it down with a series of fine abrasive paper. Despite what looks like sketchy hand work, it's actually a very controlled process. You can see the gradual emergence of machining marks, then flattening to a constant metal surface with no blue remaining, then finish with polish rubbers so you can see a reflection. On the same valve cage I watched the vacuum times increase from horrible 3-5 seconds to 30+secs, and which point I call it good.