Added code to handle Chianti 2016 collisional excitation rates (data type 98) and also ad hoc treatment of inner shell collisional ionization (data type 97) from Patrick Palmerit fac calculations for Fe XXIV. Added new function upsiln used in this calculation. Fixed error in lower level statistical weight calculation for data type 95 (Bryans collisional ionization). Increased dimension of dummy array used in reading in atomic data in readtbl. This was filling and causing erroneous results for highest Z elements (Cu, Zn). Change to main xstar routine to prevent writing detailed step-by-step data unless write switch is set or unless npass>1.