A quick update. It turned out to be a bad tube. Least likely suspect!

Had a coupla big 100K resistors, somewhat close to tolerance and I tried them both swapping out, still same results...

It was frustrating because it would take at least 5 minutes for things to stabilize. But while testing I noted some noise on the circut. The meter was bouncing as if a signal. I isolated all rf input and it still flicked around at times flicking to S7 or so and staying there for a half a second or so.
Well the long warm up time got me to thinking. I dug into the tube box found 3 6AU6 tubes and played round robbin with three tubes in two slots. It was slow going at first given a min of 15 min warm up but eventually the flakey tube betrayed itself. .

Anyhow its rock solid now. Thanks all!