Allow units up to 999 to support processing of ENDF/B-VIII.1 light water TSL evaluation

[paulromano]

Currently, unit numbers for files are limited to be less than 100, which as best as I can tell was a limitation in Fortran prior to the Fortran 90 standard (and/or support in compilers). In modern Fortran, there is no such limitation and thus NJOY should allow higher unit numbers if and when needed. I found that when processing the ENDF/B-VIII.1 evaluation for H in H2O, I needed unit numbers >= 100 because there are so many temperatures listed. In my NJOY input, I run ACER once for each temperature and ended up needing unit numbers all the way up to 213. The change in this PR simply allows NJOY to use unit numbers up to 999, which should cover most conceivable use cases.

[whaeck]

Yes, for H in H2O there's 94 temperatures so if you do everything in a single NJOY run, you will run out of numbers. I tend to run a single input file for each temperature to get around any issues.

However, if this increases the tape number count without issues, I'm all for it. We may have to verify that there's no other places where the tape number is checked outside of the openz function.

We will also need the following:

• an update to the release notes under NJOY2016.78
• a test using tape numbers larger than 100. This can be any input deck, but is should run relatively quickly.

[paulromano]

@whaeck I just added a short test and updated the release notes. Let me know if anything else is needed!