Remove Manifest.toml (#3783)

* Remove Manifest

* Move some imports to global dependencies

* Use more precompile tasks

* Add PKG

* Make GPU visible to tartarus init plus proper init

* Move imports to dependencies_for_runtests

* Rm using KA

* Clean up + find a way to define test utils only once

* Testing something

* Remove unused test dependencies

* Reorganize dependencies

* Give it one more try

* Rm TestEnv

* Auto retry distributed tests

* Update distributed pipeline

.buildkite/distributed/pipeline.yml

@@ -3,34 +3,26 @@ agents:
   slurm_mem: 8G
   modules: climacommon/2024_05_27
 
-
 env:
   JULIA_LOAD_PATH: "${JULIA_LOAD_PATH}:${BUILDKITE_BUILD_CHECKOUT_PATH}/.buildkite/distributed"
   OPENBLAS_NUM_THREADS: 1
+  JULIA_PKG_SERVER_REGISTRY_PREFERENCE: eager
+  JULIA_NUM_PRECOMPILE_TASKS: 8
+  JULIA_NUM_THREADS: 8
   OMPI_MCA_opal_warn_on_missing_libcuda: 0
 
 steps:
   - label: "initialize"
     key: "init_central"
+    env:
+      TEST_GROUP: "init"
     command:
       - echo "--- Instantiate project"
-      - "julia --project -e 'using Pkg; Pkg.instantiate(;verbose=true); Pkg.precompile(;strict=true)'"
-      # force the initialization of the CUDA runtime as it is lazily loaded by default
-      - "julia --project -e 'using CUDA; CUDA.precompile_runtime(); CUDA.versioninfo()'"
-      - "julia --project -e 'using MPI; MPI.versioninfo()'"
-      # Download artifacts by running an empty testgroup and thereby executing /test/runtests.jl
       - "julia -O0 --color=yes --project -e 'using Pkg; Pkg.test()'"
-
-      - echo "--- Instantiate status"
-      - "julia --project -e 'using Pkg; Pkg.status()'"
-
     agents:
       slurm_mem: 120G
       slurm_gpus: 1
       slurm_cpus_per_task: 8
-    env:
-      JULIA_NUM_PRECOMPILE_TASKS: 8
-      TEST_GROUP: "init"
 
   - wait
 
@@ -54,6 +46,11 @@ steps:
       slurm_mem: 120G
       slurm_ntasks: 4
       slurm_gpus_per_task: 1
+    retry:
+      automatic:
+        - exit_status: 1      
+          limit: 1
+
 
   - label: "🦾 cpu distributed solvers tests"
     key: "distributed_solvers_cpu"
@@ -75,6 +72,10 @@ steps:
       slurm_mem: 120G
       slurm_ntasks: 4
       slurm_gpus_per_task: 1
+    retry:
+      automatic:
+        - exit_status: 1      
+          limit: 1
 
   - label: "🤺 cpu distributed hydrostatic model tests"
     key: "distributed_hydrostatic_model_cpu"
@@ -96,6 +97,10 @@ steps:
       slurm_mem: 120G
       slurm_ntasks: 4
       slurm_gpus_per_task: 1
+    retry:
+      automatic:
+        - exit_status: 1      
+          limit: 1
 
   - label: "🦍 cpu distributed nonhydrostatic regression"
     key: "distributed_nonhydrostatic_regression_cpu"
@@ -117,6 +122,10 @@ steps:
       slurm_mem: 120G
       slurm_ntasks: 4
       slurm_gpus_per_task: 1
+    retry:
+      automatic:
+        - exit_status: 1      
+          limit: 1
 
   - wait
 
@@ -125,3 +134,4 @@ steps:
       - "build_history staging"
     artifact_paths:
       - "build_history.html"
+