The "tic" size of the ICS Treasury Spread is published by the CBOT as the front leg (smallest tic increment) product. The quotation feature (net change from previous days settlement) is an artifact from the days of floor trading. That's a convenient way for CBOT to arrange DOM pricing, but it's a misnomer in that the only way that's a CONSTANT tic size over time would be if you were spreading, let's say, 30 yr Bonds vs. UltraBonds where both products have an identical point value and tic value 1/32nd. Conversely, the NoB will appear priced at one half of 1/32nd - but as you've pointed out, that Spread will shake and fidget around intraday like a whore in church - no way that Bond 32'nd gets split up.
The CBOT ICS Spread was designed as a dandy fill mechanism, not as a unique standalone contract per se. They want it to be indivisible by design (you can offset it by other means).
As you know, every quarter the CBOT publishes a fresh Spread Ratio .pdf spec sheet because the OTR CTD cash changes (especially for the shorter expires). If you take that sheet, built a synthetic Spread expression, and then overlay the CBOT Spread exchange expression on the same chart you will find that they do indeed overlap. So, the good news is, let's say you were bearish the FYT over let's say a week period it would work out exactly as charted.
How do I handle this idiosyncrasy ? When my clients are paper trading, I tell them to mark each leg according to the daily settlement, calculate the P&L differential from their original fill prices, and mark that Spread on their tracking sheets accordingly. Same holds true for other intermarket spreads with different tic valuations.
This also explains why a fair number of my clients just stick with intramarket spreads.