lag_energy_spectrum: Add freq.-resolved RMS calculation

- The RMS in each frequency band for each energy bin is now calculated
  and written into the output file
- Calculation is the same as in foucalc. However, there might be a bug
  in foucalc, namely the RMS frequency filtering in foucalc selects
  the range [fmin,fmax] but it should be ]fmin,fmax] as otherwise lower
  bins are counted twice when looped over multiple frequencies. In the
  lag_energy_spectrum function, the filtering is done on ]fmin,fmax].
  This likely has no influence on the RMS calculation as the DFT grid
  is most likely different from the user's frequency grid.
- Write n_segments, n_frequencies, f_min_dft, f_max_dft as keywords
  instead of columns because they must be the same for each given
  energy-resolved lightcurve (add check which throws a warning if this
  is not the case)