I guess my method is eating up a lot of ram or something.
Well, loading sound files into memory is going to use memory, but that's not what the message is about.
FWIW, 441000 32-bit samples = 441000 * 4 bytes = 1764000 = about 1.7 MB, so a few such files would consume a few MB. In modern computers, this is nothing to be concerned about.
I believe the warning relates to the "save contents" checkbox in the array's properties. If enabled, it will write the array's data into the pd patch file... in ASCII decimal format, 9-10 bytes per sample. This is really inefficient. Usually you don't want to do this for audio arrays. So it tells you, "hey btw, you're getting piles and piles of audio data written out as text into your patch, are you sure?" and then you have the opportunity to right-click on the array and uncheck the "save contents" box.
Usually you shouldn't see that, though, because soundfiler should uncheck that box for you when you do "read -resize":
"warning: array array2: clearing save-in-patch flag"
So I'm not sure how you're ending up with the box checked (unless you saw the "clearing save-in-patch" warning and manually went to Properties to re-check it).
Anyway if you're loading audio from a disk file, then there's no point in saving it in the patch. Save-in-patch is good for small arrays where you don't want to have a separate file for the data... but you already have a separate audio file. Hence "save-in-patch" + soundfiler is only wasting CPU time and disk space.