Sync HAFS submodules with their corresponding authoritative branches as of 07/23/2021

[BinLiu-NOAA]

Sync HAFS submodules with their corresponding authoritative branches:

• hafs_forecast.fd as of 08/05/2021
• hafs_gsi.fd as of 08/06/2021 plus the dual-resolution 3DEnVar bug fix
• hafs_post.fd as of 08/02/2021
• hafs_utils.fd as of as of 07/23/2021
• hafs_graphics.fd/hrd_gplot as of 07/27/2021
• hafs_graphics.fd/emc_graphics as of 08/08/2021

Besides application level changes were made accordingly with the updated submodules.

This PR addresses issue #80.

Notes:

• The bug of wrong Time dimension in FV3 restart sfc_data files has been fixed in ufs-weather-model through this ufs-weather-model PR.
• The bug fix in for the dual-resolution EnVar analysis in GSI (hafs_gsi.fd) contributed from OU collaborators has also been included in this PR. With that the HAFS ENSDA configurations can now work properly.
• As for the hafs_forecast.fd submodule, support/HAFS branch is identical to the ufs-weather-model develop branch as of 08/05/2021. More information can be found through this PR (Merge HAFS coupling related developments back to ufs-weather-model ufs-community/ufs-weather-model#715)

[BinLiu-NOAA]

Currently, the ufs-weather-model level standard regression tests went through on Orion. As for the HAFS application level regression tests, all the regional configurations went through. But the two global-nesting tests failed with the forecast job, which will need to be resolved before this PR can go back to HAFS develop.

[BinLiu-NOAA]

A new PR #86 is used to replace this PR. Closing this draft PR #82.