I was pretty certain @David's comment was spot on, that this was some one-use funny acronym from just this paper, but a search of Google Scholar suggests that it has indeed been propagated sufficiently to have meaning beyond this paper, while also originating, as these things do, in some clumsy one-off usage.
One of them gives a clear definition that is consistent with what you supposed:
The acronym FIIND (domain with function to find) was coined several years ago to indicate a highly conserved protein region with unknown function[4]
D'Osualdo, A., Weichenberger, C. X., Wagner, R. N., Godzik, A., Wooley, J., & Reed, J. C. (2011). CARD8 and NLRP1 undergo autoproteolytic processing through a ZU5-like domain. PloS one, 6(11).
The paper they cite (reference 4 in the previous quote) is in fact this one:
Tschopp, J., Martinon, F., & Burns, K. (2003). NALPs: a novel protein family involved in inflammation. Nature reviews Molecular cell biology, 4(2), 95-104.
...which probably looks familiar since it is in fact the paper you started from. Importantly, though, this is not a generic term for a domain with unknown function, it's a specific domain which was known to be highly conserved (and since found in several other proteins) - it's just that it's specific function was at the time unknown.
This paper is a recent review of NLRP1 that may also be useful and has some discussions of what has been found of the functions of the FIIND:
Mitchell, P. S., Sandstrom, A., & Vance, R. E. (2019). The NLRP1 inflammasome: new mechanistic insights and unresolved mysteries. Current opinion in immunology, 60, 37-45.
