A quick fix for incorrect usage of real(i_kind) in mg_input.f90

[TingLei-daprediction]

Description
A quick fix for incorrect usage of real(i_kind) in mg_input.f90 , which was identified by D. Kokron.
Fixes #757
Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?
The building succeeded. And the changed code mg_input.f90 is not used by other codes in the current GSI and hence the change hadn't and would not affect any GSI runs.

Coauthor : D. Kokron

[RussTreadon-NOAA]

@TingLei-NOAA, @hu5970 , and @ShunLiu-NOAA: I did not realize that we do not exercise mgbf in any of the regional ctests. This is not good. We intend to use mgbf in an operational realizations of the GSI, right? If true, we should exercise mgbf code as part of our standard GSI regression (ctest) suite.

What is the impact of correctly defining the variables in question as integer(i_kind) instead of real(i_kind)? Someone needs to run a test to quantify the impact, if any, of this bug fix.

[ShunLiu-NOAA]

@RussTreadon-NOAA, we have no plan to use MGBF in RRFSv1. We expect to use it in RRFSv2 with JEDI.

[RussTreadon-NOAA]

Thank you @ShunLiu-NOAA for this very surprising news. If we do not intend to use mbgf in an operational realization of GSI why did we add it to the GSI repository?

[ShunLiu-NOAA]

I am not sure if 3DRTMA will implement this. Also, for JEDI-MGBF development, it is necessary to test MGBF with GSI first. This is what Ting and other developers are working on.

[RussTreadon-NOAA]

Thanks, @ShunLiu-NOAA. Given your reply we need a mgbf ctest in GSI. This PR can use this new ctest. Who will create the mgbf ctest?

Adding @ManuelPondeca-NOAA for awareness.

[ShunLiu-NOAA]

@RussTreadon-NOAA Thank you. We will work with Ting to provide a regional DA test case.

[RussTreadon-NOAA]

Great! Thank you @ShunLiu-NOAA and @TingLei-NOAA .