What about a good old spring washer ?
It will be unlikely to produce enough force.
Still thinking of the soldering. Not that confused though. The soldering temps for these are 250C though I can say for sure how much difference there is between the tip and dial reading.
Usually I tin both surfaces separately and then join by applying the tip. If I did that it shouldn't take long to make the connection. Am I seriously better off using single core wire though?
Soldering directly to the leads is not problematical as far as temperature. Like Brutte says, the problem is a long term reliability issue with the connection itself; in a high vibration environment, either the lead to package interface or the solid to stranded wire interface created by solder wicking tends to break.
I have lots of old equipment which has TO-3 and TO-220 parts mounted on a heat sink away from the printed circuit board and connected with stranded wire leads soldered directly to the pins of the part and there has not been a problem. If you do this, include strain relief near the soldered junctions to minimize bending. I usually cut off the narrow portion of the TO-220 lead when doing this but this can make things worse.
"Anti wicking tweezers" are available for holding stranded wire in place during soldering which prevent wicking of the solder up under the insulation.
For what it is worth, they make sockets for TO-3 and TO-220 parts and I do not think they are more reliable than soldering wires to the leads. TO-3 parts soldered directly to boards can have major problems with thermal expansion causing enough force to break the solder joints. I imagine that TO-220 parts might have the same problem if mounted vertically.