Save marine ensemble mean in the ensemble recentering app

[shlyaeva]

For the diagnostics purposes it would be very useful to have ensemble mean for ocean and ice at every cycle (currently we can only infer them from restarts archived ~ once a week). This would also allow to compute recentering increment at every cycle for diagnostics purposes. Note: we already save ensemble variance for ocean and ice at every cycle.

Anticipated changes are:

1. adding an ensemble mean output section in soca_ens_handler.yaml.j2 similar to:
   

   GDASApp/parm/jcb-gdas/algorithm/marine/soca_ens_handler.yaml.j2

   Lines 106 to 110 in b52d099

   	ensemble variance output:
   	datadir: ./
   	date: "{{ marine_window_end_iso }}"
   	exp: ensvar
   	type: incr

2. copying the mean files to comroot, similar to how variance is copied in:
   

   GDASApp/parm/marine/marine_ecen_config.yaml.j2

   Lines 86 to 87 in b52d099

   	- ['{{ DATA }}/ocn.ensvar.incr.{{ WINDOW_END | to_isotime }}.nc', '{{ COM_OCEAN_ANALYSIS_TMPL | replace_tmpl(tmpl_dict) }}/{{ APREFIX_ENS }}bg_ensvar.nc']
   	- ['{{ DATA }}/ice.ensvar.incr.{{ WINDOW_END | to_isotime }}.nc', '{{ COM_ICE_ANALYSIS_TMPL | replace_tmpl(tmpl_dict) }}/{{ APREFIX_ENS }}bg_ensvar.nc']

3. archiving the ensemble mean files similar to variance files as in:
   https://github.com/NOAA-EMC/global-workflow/blob/250b0130f0007ae8c70d9db5bde9d944899510f4/parm/archive/enkf.yaml.j2#L152-L153