[RM-8005] [linemodel] do not switch to nocontinuum or fromspectrum in second pass continuum refit

Created on 2023-04-06 09:14:09 by Didier Vibert. % Done: 50

when the continuum is refitted, eventually partially in the case @ LineModelSolve.linemodel.secondpass.continuumfit = "refitfirstpass"@,
the amplitude can be found compatible with zero (resp. negative) in the considered second pass redshift window, whereas it was found in the 1st pass full z range, that at some other redshift the continuum amplitude is significantly not null (resp. positive). Currently, the code switch to "nocontinuum" (resp. "fromspectrum") globally , for all 2nd pass candidates.

We should either make the switch for only the candidate where it appears, or not switch at all, since in first pass the continuum was found significant.
I vote for 2nd option. Moreover the whole pdf, outside the 2nd pass windows, was computed using continuum template fitting, thus we should keep this info and all the associated parameters (name of template, amplitude, ism, igm)