The best method to remove the anodizing is a carbide cutter on the end of a boring bar. It's pretty tough.
If you don't have a lathe, you're down to sanding it out. Oven cleaner will work on decorative anodizing but doesn't really touch hard anodizing. Plus you risk ruining the exterior that you want to keep.
If you can find one, get one of those expanding rubber cylinder arbors that sanding sleeves slip over. Mount it in a drill press and run it up and down. Those arbors usually have short shanks. I have an extension that I made out of a Milwaukee spade bit extender that's cut down to about 4" long. Don't use a drum much more than half the diameter of the tube ID. Any larger and it will begin orbiting the ID rather than rotating against it. The drill press isn't moving so you are. If it doesn't drop away freely when tilted at an angle, it's likely going to catch, the force will cause the other end do orbit in a larger radius and the tube will break a few finger. Or worse. you can accomplish the same with a drill in a vice or clamped to a table top.
Your best bet is a press fit. The surfaces should be as smooth as possible. If not, you have tiny contact points and alot of air. Even a tiny fraction of a MM is an effective insulator, so the better the contact, the better the heat conductivity. Thermal paste is better than not if you can't get a good interference fit. It's going to be tough to get uniform application with a tight fit so you're back to doing what you can to get the best metal to metal fit possible. The alternative would be to have a larger gap and fill it with a thermal adhesive.
You have to determine how far you want to go and that depends on what you're trying to accomplish. If you're using an XML in a Mag and have an inch-tall heat sink, remove the anodizing and goop it up. If you're putting 10 amps into an SST90 in that mag, find a good machinist.
You can do the basic math for each material interface, beginning at the diode and ending at the exterior of the tube. The numbers are only going to be relative but the exercise will show you where the thermal bottlenecks are. Unless you do that, you're just guessing- and maybe wasting time, effort and money. Nothing is worse than investing in something you enjoy and not being happy with the results. And it's very avoidable. It's a hobby- it should be fun.
