Your actual formula in specific amounts would help for trouble shooting. Which palm are you using? Palm oil? PKO? PKO flakes, hydrogenated? *It matters, as their fatty acid profiles are different and they contribute to trace differently. What is your SF set at? How much of a water discount are you using? (Or lye concentration, which I prefer to use, but they amount to the same thing.) Also, are you sure your scale is working correctly? Have you calibrated it recently? If not, I would suggest you do that to ensure your measurements are accurate.
Check out this
link for a list of the fatty acids and how they contribute to trace. I have found that by subbing RBO for some other soft oils has lead to an unexpectedly faster thickening of the soap. It appears that others have found this to be the case also; I suspect it is the amount of palmitic already in the recipe and how much RBO is also in the recipe (RBO is 22% palmitic).
As far as false trace, I just don't know with the given information. It could also be my experience level. I have experienced separating in the mold, and the kind of results you described. I re-batched to save the soap. But I don't recall if I was soaping as warm as you were, nor have I looked back at the formulas used when that happened.
And when you say, 'a variety of fragrance oils' it would really be helpful to know which ones were used in the failing batches. Include the vendor name as well as the actual amount added. That can give a real clue to the problem. Have you looked into each of the FO's used to see if they cause issues?
Here is one resource for looking up fragrances. Another source is the actual vendor; if they don't have the information available on their website, then come here and ask for feedback regarding particular vendors.
* Example: PKO flakes, hydrogenated are 90% lauric + myristic + palmitic + stearic, whereas plain PKO is 75%, and Palm Oil is merely 50%. Castor is 90% Ricinoleic, which also leads to accelerated trace. RBO has 22% palmitic, even though it is a soft oil. And of course CO has 79% of the top 4 fatty acids that contribute to acceleration and hardness.