reconr, MLBW instance of no data is mishandled and tries to divide by zero

[tegaines]

For resolved resonances, MLBW. In a resonance block where NLS=3, one of the values of L has no data associated with it so NRS=0. NJOY then does a divide by zero when working out the ncyc for the number of list entries there will be which ends as expected.

Wherever or not there should be data or not, there appears to be no catch that a zero might be there. Unknown if having no data is against ENDF-6 format since it doesn't say you can't... If I make up a fake bit of data and set NRS to 1, with 6 words the file now builds.

Applies to JEFF-4T4.
83-bi-210g.txt
18-ar-37g.txt
Ar37 and Bi210 both show this problem.

[tegaines]

If I also set NLS=2 and just delete the L=1 line it also processes successfully. (So just L=0 and 2 are given)