Really, launching rockets into tornadoes to collect data is not entirely without merit. No to say that I'm endorsing the method this group proposes, but fundamentally this is could be a sound technique to gather data. Instruments that collect temperature, pressure and moisture data are extremely small these days thanks to advances in electronics - so you certainly can make meaningful measurements that would fit on a rocket - and the financial costs for what ends up becoming a disposable weather instrument become more realistic when the cost can be reduced to a few hundred dollars per probe (rockets are fairly cheap).
There are designs underway to instrument small planes to fly around storms during the next VORTEX project, assuming it gets funded, but these instruments are not meant to be flown into the actual tornado. As previously mentioned, there were some efforts by Stirling Colgate to launch airplane based rockets into tornadoes back in the 80's, but these efforts were largely unsuccessful. Tim's probes are extremely valuable as they are sampling conditions at the ground - where damage occurs - but alone don't provided enough information to fully understand the dynamics of tornadoes.
Another use for rockets proposed in the past is to use them to disperse chaff (small pieces of foil) which allows radars to then "see" air motions in portions of a storm where the chaff is deployed - which is useful for sensing the winds in portions of a storm where this is no precipitation to make it "visible" to the radar.
Erik Rasmussen, who is an innovative although somewhat isolated scientist, is leading the effort to develop the unmanned aircraft probes described above as a part of the RECUV Project, but has also been working with concepts to develop rocket-based deployments, and this may still someday come to fruition.
Glen