Use C++20 in both standalone and coupled builds

[andrewdnolan]

The changes to Omega's build system to support C++20, brought in by #405, only affected standalone builds. After #566, we are using Kokkos5 and need C++20 in coupled mode (in addition to standalone). This PR moves setting the CMAKE_CXX_STANDARD and CMAKE_CXX_STANDARD_REQUIRED into the common macro, so that they apply to both couple and standalone builds.

Without the changes in this PR, any C/G cases that attempted to build off develop would produce and error along the lines of:

-- Configured git information in /lustre/orion/cli115/proj-shared/anolan/e3sm_scratch/SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu.develop/bld/cmake-bld/externals/kokkos/generated/Kokkos_Version_Info.cpp
-- Kokkos is configured for CMake languages CXX compilation (using GNU version 14.2.0)
CMake Error at /autofs/nccs-svm1_home2/anolan/repos/Omega/develop/externals/ekat/extern/kokkos/cmake/kokkos_test_cxx_std.cmake:89 (message):
  Kokkos requires C++20 or newer but requested 17!

Checklist

• Linting

  • Pre-commit run locally

• Building

  • CMake build does not produce any new warnings from changes in this PR

• Testing

  aurora, oneapi-ifx, mpich

  • CTests Pass
  • Polaris omega_pr Pass

  chrysalis, oneapi-ifx, openmpi

  • CTests Pass
  • Polaris omega_pr Pass

  frontier, craygnu, mpich

  • SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF

  frontier, craygnu-mphipcc, mpich

  • CTests Pass
  • Polaris omega_pr Pass

  pm-cpu, intel, mpich

  • SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF

  pm-gpu, gnugpu, mpich

  • CTests Pass
  • Polaris omega_pr Pass

[sbrus89]

@mwarusz, would you mind taking a look? Thanks!

[andrewdnolan]

Testing: frontier, craygnu

With the changes in this PR, I can successfully build SMS_Vmct.T62_oQU240.GOMEGA-IAF on frontier with the craygnu compiler. I'm having issue running that case, because the LCRC server is down as part of the maintenance. The input data for the case is missing, which is why CIME is attempting to connect to the LCRC server.

I'm currently setting up tests on perlmutter, which should hopefully have the needed inputdata downloaded.

[andrewdnolan]

Testing: pm-cpu, intel

I ran: ./create_test --queue debug --no-run --machine pm-cpu SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF . Then edited the user_nl_omega to contain:

Omega:
  Tendencies:
    KPPNonLocalTracerFluxTendencyEnable: false
  VertMix:
    KPP:
      Enable: false

and everything passed:

PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel CREATE_NEWCASE
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel XML
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel SETUP
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel SHAREDLIB_BUILD time=159
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel MODEL_BUILD time=313
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel SUBMIT
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel RUN time=85
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel MEMLEAK insufficient data for memleak test
PASS SMS_Vmct_Ld1.T62_oQU240.GOMEGA-IAF.pm-cpu_intel SHORT_TERM_ARCHIVER

---

With KPP on, I get a failure at the beginning of the second coupling interval. @vanroekel has a PR coming up to fix whatever is wrong with KPP.

[andrewdnolan]

Testing, frontier, craygnu

Running:

./create_test --no-run --compiler craygnu SMS_Vmct.T62_oQU240.GOMEGA-IAF

with the same user_nl_omega as above, which disables KPP, enables the case to pass:

PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu CREATE_NEWCASE
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu XML
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu SETUP
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu SHAREDLIB_BUILD time=85
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu MODEL_BUILD time=241
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu SUBMIT
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu RUN time=48
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu MEMLEAK
PASS SMS_Vmct.T62_oQU240.GOMEGA-IAF.frontier_craygnu SHORT_TERM_ARCHIVER

[mwarusz]

Approving based on visual inspection and @andrewdnolan testing.