[RM-6103] Study the effect of secondpass.halfwindowsize on final results and on time gain, for PFS and Euclid

Created on 2020-11-05 10:47:13 by Mira Sarkis. % Done: 0

• Parameter in question: linemodelsolve.linemodel.secondpass.halfwindowsize (added in #6057)
• Conduct tests on :
  • PFS dataset (e.g., PFS7 dataset??)
  • Euclid dataset (e.g., Euclid_Tiger dataset??)
• Use plots of #6051 to decide on the values to test:
  • a first guess: 2E-3 for Euclid et 1E-3 for PFS
• Metrics to study:
  • Time Gain, i.e., effect of secondpass.halfwindowsize on secondpass duration, especially for PFS
  • Success Rate in function of different values
• Expected outcome, for Euclid and PFS:
  • Determine secondpass.halfwindowsize value as a good tradeoff between SuccessRate and time gain.
  • Accordingly update the default param.json for Euclid and PFS