I know what you mean about methods being locked down. Sometimes its easier to make a complicated calculation that it is to change an established, inefficient method!
Reading your post, I suspect the CFs only generate after processing a result set because it requires the calibration curves be generated in full before you apply any CFs to them, so any CFs with Amount field in them can be tricky, depending on how your brackets are set up. I know for instance that creating multiple calibration curves in a sample set will affect your CFs if Amount is included, which would point to why only reprocessing a result set would work.
A few things to check, like your search order is Result Set Only, and sample/peak type set to All.
If that's ok, try naming your CFs in Alphabetical Order- summary CFs in particular require this to operate in the correct and logical order. So taking your first CF call it something like A_CF_One (just make sure A is first letter), and formula is (C7.%.(Response)+C8.%.(Response))/2.
Call your second CF B_CF_Two and If I understand you, you are using the result for CF1 as correction factor for your samples? So if you label each sample as U followed by something, eg U1, U2, U0101 etc etc then the formula for B_CF_Two (keeping the same search order etc as first one) is:
U%.%.(Amount)*A_CF_One. Then if you need to further adjust this for LOD, you could use the CConst1 field of your processing method as this LOD value. I presume your are checking B_CF_Two against this value so the 3rd CF of C_CF_Three could be a peak bool CF with formula like GTE(B_CF_Two,CConst1) and return either text like Pass/Fail or a field itself if you pick Use as Field.
Hope this helps a bit.