restructure libomnitrace + tasking and omnitrace-causal updates (#237)

* restructured libomnitrace

- this is necessary to incorporate some of the binary analysis capabilities into omnitrace exe
- created libomnitrace-core (static)
- created libomnitrace-binary (static)
- created libomnitrace (static)
- omnitrace-avail links to libomnitrace.a
- omnitrace-critical-trace links to libomnitrace.a
- tweaked the testing
  - reduced verbosity on some of MPI tests
  - excluded trace-time-window from tests on Ubuntu 18.04
  - reduced causal e2e iterations
- minor tweak to tasking
  - manually create `PTL::UserTaskQueue` instance instead of relying on `PTL::ThreadPool` to create it

* Update formatting workflow

- source formatting uses ubuntu-22.04
- check-includes doesn't generate false positive for 'include "timemory.hpp"'

* omnitrace-causal --generate-configs

- fix config generation in omnitrace causal
- add test for omnitrace-causal + generating configs

* Fix omnitrace-object-library build

- accidentally included rocm sources in non-rocm builds

* Fix rocm compilation w/o rocprofiler

* update timemory submodule with mpi_get warning messages

* sampling offload file updates

- more verbose messages
- disable offload before stopping

* testing updates

- increase causal e2e iterations to 12
- increase lock_environment verbose to 2 (for sampling offload messages)
- fix return for omnitrace_add_validation_test

source/lib/omnitrace/library/rocm.cpp

@@ -21,12 +21,12 @@
 // SOFTWARE.
 
 #include "library/rocm.hpp"
+#include "core/config.hpp"
+#include "core/debug.hpp"
+#include "core/dynamic_library.hpp"
+#include "core/gpu.hpp"
 #include "library/components/rocprofiler.hpp"
 #include "library/components/roctracer.hpp"
-#include "library/config.hpp"
-#include "library/debug.hpp"
-#include "library/dynamic_library.hpp"
-#include "library/gpu.hpp"
 #include "library/rocm/hsa_rsrc_factory.hpp"
 #include "library/rocm_smi.hpp"
 #include "library/rocprofiler.hpp"
@@ -81,10 +81,11 @@ bool       on_load_trace = (get_env<int>("ROCP_ONLOAD_TRACE", 0) > 0);
 }  // namespace rocm
 }  // namespace omnitrace
 
+#if defined(OMNITRACE_USE_ROCPROFILER) && OMNITRACE_USE_ROCPROFILER > 0
 std::ostream&
 operator<<(std::ostream& _os, const rocprofiler_settings_t& _v)
 {
-#define ROCPROF_SETTING_FIELD_STR(NAME) JOIN('=', #NAME, _v.NAME)
+#    define ROCPROF_SETTING_FIELD_STR(NAME) JOIN('=', #    NAME, _v.NAME)
 
     _os << JOIN(
         ", ", ROCPROF_SETTING_FIELD_STR(intercept_mode),
@@ -97,6 +98,7 @@ operator<<(std::ostream& _os, const rocprofiler_settings_t& _v)
         ROCPROF_SETTING_FIELD_STR(obj_dumping));
     return _os;
 }
+#endif
 
 // HSA-runtime tool on-load method
 extern "C"