Replace mersenne twister in CALWXT_BOURG.f with standard Fortran RNG

[DusanJovic-NOAA]

In applications like ufs, where different components are built using different precision of default reals, Mersenne twister from w3emc (which does not provide overloaded interfaces) can not be used. Replace it with standard Fortran RNG.

[WenMeng-NOAA]

@DusanJovic-NOAA Can you sync your branch with upstream/develop? After that, there might be additional changes in
sorc/ncep_post.fd/TIMEF.f due to "implicit none" implemented from the last commit. See the changes in my PR #243
https://github.com/NOAA-EMC/EMC_post/pull/243/files#diff-855a3040248e9ead99fee9508ae601a579e01a9aaf7a9cc9cc7b1ac051af0ee4

[WenMeng-NOAA]

@DusanJovic-NOAA
There is a residue bug from last UPP commit, Could you add the changes in sorc/ncep_post/TIMEF.f in your branch as:

1. change
   real8 timef
   into
   real timef, etime
2. change
   real8 rtc
   into
   real8 rtc,etime

Thanks!

[DusanJovic-NOAA]

@DusanJovic-NOAA
There is a residue bug from last UPP commit, Could you add the changes in sorc/ncep_post/TIMEF.f in your branch as:

1. change
   real_8 timef
   into
   real_ timef, etime
2. change
   real_8 rtc
   into
   real_8 rtc,etime

Thanks!

I can add this. But ETIME is not standard Fortran function. Try to use only standard functions, for example SYSTEM_CLOCK().

[WenMeng-NOAA]

@DusanJovic-NOAA I might think TIMEF.f is an UPP legacy code. Could you add the upgrade in your branch too? Thanks!

[WenMeng-NOAA]

@DusanJovic-NOAA Will you update the file "VERSION" from 10.0.0 to 10.0.1?

[junwang-noaa]

The code changes look good to me, but we do not have any testing on the Fortran random number generator, so the impact of the changes to post products is unknown. As Hui-ya pointed out, a new method may come soon, so this may be some temporary fix.

[WenMeng-NOAA]

The UPP regression tests for GFS, GEFS, FV3/LAM, RAP, HRRR were conducted on WCOSS dell, cray and Hera. The precipitation types related fields will have changed results in GFS, GEFS, FV3/LAM products. With this PR merged in develop branch, if there might be scientific concerns from the developing model parallel run, I will provide model developers the guidance of switching back to mersenne twister in their standalone post run.