Allow initial ledgers to use a preordained timestamp

[geordie666]

This PR adds an option to pass a timestamp when making initial ledgers. The passed timestamp will then be used in place of the TIMESTAMP column generated by make_mtl(). MWS_FAINT targets can be exempted from this forced timestamp, and will instead adopt the usual "recent" timestamp generated by make_mtl().

The purpose of these changes is to address the issue where 1.0.0 ledgers (and target files) were generated without MWS_FAINT targets, as fixed in #733. This PR should allow 1.1.0 ledgers to be generated with TIMESTAMPs that correspond to 1.0.0 ledgers for all targets except MWS_FAINT, as suggested by @sbailey.

Some other minor updates are included in this PR:

• The data model has been altered for the ledgers:
  • ZS and ZINFO are replaced by Z_QN, as QuasarNP redshifts are the only non-redrock information that we'll be using to make LyA decisions.
  • This data-model-update is not backwards compatible. So, if you're using desitarget functionality to read the never-to-be-spoken-of-again 1.0.0 ledgers, then you'll have to use version 1.0.0 of desitarget.
• A bug where an extraneous hpxlist was passed to write_secondary(), that was introduced in #734, and was noticed by Anand has been fixed.

[geordie666]

A few example ledgers that include timestamps that have been "forced" for every target class except MWS_FAINT are available at NERSC in /global/cscratch1/sd/adamyers/PR739.

[coveralls]

Coverage decreased (-0.1%) to 58.975% when pulling 2f0c7a0981511e03565632acddf65e27ff31f52d on ADM-hack-init-MTL into f2c6d4d2f4def25f6c79e1dfe235839640e73d9b on master.

[geordie666]

@sbailey: I think this is ready to merge, and I'd like to merge it in about an hour so I can tackle the MTL loop after lunch without worrying about creating merge conflicts.

[sbailey]

Thanks for the explanation, looks good. Merging now.