JMHO, but I don't see yesterday as a clear example of NWP failure. Virtually all CAM output I looked at in the 48 hours leading up to the "event" showed extremely messy, largely linear convection right along the front sweeping across the area. UH, while sometimes decent in magnitude, mostly had a telltale messy character as well. It wouldn't have surprised me to see a few more QLCS tornadoes than we actually did, but it isn't at all surprising that there were no daytime tornadoes or tornadoes from classic supercells.
Aside from CAMs, the warm sector environment depicted by convection-parameterizing models ranged from very unimpressive to moderately impressive, with the more impressive end of that spectrum generally represented by models that underdid mixing and were too late in depicting widespread convection (e.g., yesterday morning's RAP runs).
If there's one area where "guidance" deserved blame, it would have to be post-processing techniques like Nadocast and SPC's calibrated HREF-based suite of severe probabilities. All these methods produced 10-15% tornado contours, and Nadocast included a hatched area. One weakness of these methods is that HREF UH is an important input to them (as it should be), but when large UH is produced by convection right along or even just behind a surging front or other boundary, the algorithms sometimes fail to "appreciate" fully that this UH is detached from the favorable warm sector environment just to its south and east. In other words, it's easy for us to look at a HRRR run and recognize manually that the big UH streaks in KS yesterday were from storms likely getting undercut, but some of these algorithms may give those UH streaks at least partial credit as tornado threats for being so close to favorable STP values say 10-30 miles to their east.
On a less NWP-specific note, having now chased based out of central OK for almost 20 years, it's humbling to realize how many of these characteristic "big trough and good traditional parameters, but messy and slightly veered out" busts or semi-busts we continue experiencing near and just W of the I-35 corridor in OK. I recall multiple events pretty similar to this, with similar forecasts, as an undergraduate in the mid-late 2000s. Fast forward to this year, and we've had several of these, including yesterday. I must admit, if you told me back then that there'd be relatively minimal improvement in the false alarm rate for this regime over the next 20 years (and even that cutting-edge AI/ML methods would highlight them as big events), it would've surprised me.