Make rcfg_cloudsat a HEAP variable

[bartgol]

The rcfg_cloudsat variable can be large. If the cospIN object is on the stack, this can cause lots of stack memory, which can fail unless ppl run ulimit -s unlimited first. In particular, cospIN was 28MB, while the common stack size is 8MB.

[brhillman]

@alejandrobodas, FYI, this is causing us issues with running in E3SM/SCREAM. We have been using a local version of the code with this change, but are trying to update to the latest version of the repo so it would be nice to get this in so we can actually track COSP2 development.

[alejandrobodas]

@bartgol @brhillman many thanks for this. The integrations tests fail, likely because the intel compiler has changed. I'm addressing that in #100. As soon as #100 is merged we'll make progress with this one.
I've looked at the changes and I wonder if you need a deallocate statement at the end of destroy_cospIN?

[bartgol]

Uhm, aren't allocatable variables in Fortran automatically deallocated when they go out of scope (unlike pointers which need to be manually deallocated)? I thought that was the big advantage of allocatable with respect to pointer..

[alejandrobodas]

Uhm, aren't allocatable variables in Fortran automatically deallocated when they go out of scope (unlike pointers which need to be manually deallocated)? I thought that was the big advantage of allocatable with respect to pointer..

Indeed, please disregard my comment.