Automatic pe settings

Automatically set up PE ini file depending on source parameters reported by detection pipelines.

• Change PE ini file depending on mass ratio q
  if q > 1/7: use IMRPhenomPv2 with ROQ and nested sampling else if 1/7 > q > 1/17: use SEOBNRv4ROM with ROQ since IMRPhenomPv2_ROQ can not be applied in this range
  else if 1/17 > q: use non-ROQ SEOBNRv4ROM and use nparalell=10, neff=300 and lalinferencemcmc to speed up PE.
  In addition, margphi option is always used to speed up runs.

• Move the code in lalinference_pipe (https://git.ligo.org/lscsoft/lalsuite/blob/master/lalinference/python/lalinference/lalinference_pipe_utils.py#L210-246) to determine srate, fhigh, seglen, ifos.
  After discussion with Vivien, we decided to move the code in lalinference_pipe which overwrite the settings at online PE follow-up to gwcelery . Instead we let gwcelery to upload ini file to GraceDB so that PE ROTA can understand what settings should be used for each event. Merge request for that change will be opened soon.

~~ini files generated by GWCelery are in https://ldas-jobs.ligo.caltech.edu/~soichiro.morisaki/online_pe/variational_settings/ .~~
ini files generated by fixed version of lalinference.py are in https://ldas-jobs.ligo.caltech.edu/~soichiro.morisaki/online_pe/variational_settings/updated .

cc. @vivien @john-veitch @cjhaster @leo-singer