Major cleanup of convect_deep.F90

[whannah1]

This is a continuation of the ZM overhaul and refactoring, which is limited to cosmetic cleanup of the general interface for any implemented deep convection scheme (convect_deep.F90), which is currently limited to Zhang-McFarlane.

[BFB]

[rljacob]

notes: working on testing it.

[whannah1]

The following tests pass on Chrysalis:

  e3sm_atm_developer_intel
  e3sm_atm_developer_gnu
  SMS_Ld32.ne30pg2_r05_oECv3.F2010.chrysalis_intel
  SMS_Ld32.ne30pg2_r05_oECv3.F2010.chrysalis_gnu
  SMS_Ld32.ne4pg2_oQU480.F2010.chrysalis_intel
  SMS_Ld32.ne4pg2_oQU480.F2010.chrysalis_gnu

The performance of the SMS_Ld32 tests are indentical, here's the ZM timer from SMS_Ld32.ne4pg2_oQU480.F2010.chrysalis_intel:

          name                                            on  processes  threads        count      walltotal   wallmax (proc   thrd  )   wallmin (proc   thrd  )
    base: "a:zm_conv_tend"                                 -         96      192 1.476480e+05   1.692972e+02     1.337 (    11      1)     0.591 (    73      1)
    test: "a:zm_conv_tend"                                 -         96      192 1.476480e+05   1.692972e+02     1.337 (    11      1)     0.591 (    73      1)

[singhbalwinder]

Thanks, @whannah1. Do these longer tests compare against baselines as well?

[whannah1]

Thanks, @whannah1. Do these longer tests compare against baselines as well?

Yes - I'm now creating my own baselines for all tests using a branch that exactly matches the point where each ZM feature branch starts.