Never tried it in any other orientation other than vertical. I think several reasons for this being best is that it is easier to prevent/manage air bubbles...you don't want them in your RIMS tube especially anywhere near the heating element. also if horizontal then there might be increased risk of stagnation of wort around the heating element...no evidence for this but just feels like it might be an issue.
The only experimentation I've done is to change where the thermocouple is. I started with a conventional configuration: vertical, heating element in the top half and thermocouple in the tip part. I found with this I got temperature swings and even with PID struggled to maintain steady temp. I hypothesised this was due to the thermocouple sitting right on top of the heating element and maybe picking up the temperature of the heating element rather than just the wort so change the configuration to the wort pumping through the RIMS tube into a T-piece a the top of the tube so the wort turns through 90 degrees with the thermocouple orientated horizontally in the T-piece so its not influenced by the heating element. Only used this configuration once and it worked fine, so based on a data point of one, it seems to have been a successful change.