Multi-temperature writing bug in NJOY resulted in incorrect data in "official" data libraries

[paulromano]

Just creating this issue for posterity — there was a bug in NJOY that caused our normal multi-temperature nuclear data library generation procedure to produce incorrect data for some temperatures. As that bug discusses, if you ran an NJOY deck and asked for three temperatures, say 300, 600, and 900 K, and then wrote each of them out in three separate ACE files, sometimes you would correctly get data at 300, 600, and 900 K, but sometimes you'd get 300, 600, and 600 K. This particularly affected OpenMC's automation of cross section generation because that is how we generate multitemperature data when using IncidentNeutron.from_njoy and ThermalScattering.from_njoy. Other libraries, e.g., the ACE libraries distributed by LANL, were produced such that each temperature set was produced with a separate NJOY run, so they never ran into this issue.

The NJOY developers have since fixed the bug above and it is contained in version 2016.68. I've gone ahead and regenerated our official data libraries with this version of NJOY so that the multitemperature data should be correct now (I've explicitly confirmed that the cross sections at each temperature are indeed different).