Fix bug in 'compile recombination coefficients H-like'

The command operates on either 40 or 20 shells, depending on the isoseq and species. At the time of its inception, Cloudy computed all H-like energies on the fly assuming hydrogenic scaling (1/n^2). However, it now uses NIST data, which typically do not reach n=20.

The current coding uses the number of resolved levels to distinguish between NIST and hydrogenic energies, and leads to a negative energy and a thrown ASSERT() when the command is used, because the expected data are not in the relevant Stout energies file.

With this commit, hydrogenic energies are used when NIST energy data are known, not expected, to be missing.