Radical.pilot

- binder tutorial
- expand / fix documentation and README, update policies
- docs for psij deployment
- raptor example now terminates on worker failures
- fix raptor worker registration
- fix flux startup
- fix type for `LFS_SIZE_PER_NODE`
- update HB config parameters for raptor


--------------------------------------------------------------------------------

- fix detection of failed tasks


--------------------------------------------------------------------------------

- fix type for `LFS_SIZE_PER_NODE`


--------------------------------------------------------------------------------

- pypi fix
- configurabe raptor hb timeouts


--------------------------------------------------------------------------------

- add bragg prediction example
- add initial agent scheduler documentation
- add JSRUN_ERF setup for Summit's config
- add mechanism to determine batch/non-batch RP starting
- collect task PIDs through launch- and exec-scripts
- ensure mpi4py for raptor example
- fix ERF creation for JSRUN LM (and updated tests accordingly)
- fix Popen test
- fix `Task._update` method (`description` attribute)
- fix `_get_exec` in Popen (based on provided comments)
- fix parsing PIDs procedure (based on provided comments)
- fix profiling in Popen (based on provided comments)
- fix resource manager handling in `get_resource_config`
- fix tasks handling in `prof_utils`
- fix test for launch-/exec-scripts
- follow-up on comments
- forward after scheduling
- keep `pids` dict empty if there is no ranks provided
- moved collecting EXEC_PID into exec-script
- preserve process id for tasks with `executable` mode
- switch raptor to use the agent ve
- update `metadata` within task description


--------------------------------------------------------------------------------

- AgentComponent forwards all state notifications
- document event locations
- MPI tutorial for RAPTOR
- add mpi4py to the ci requirements
- add `bulk_size` for the executing queue (for sub-agents)
- add option `--ppn` for `PALS` flavor in MPIEXEC LM
- amarel cfg
- current version requires RU v1.43
- fix Profiling tutorial (fails when executed outside from its directory)
- collect service related data in registry
- fix multi pilot example
- move agent config generation to session init
- remove obsolete Worker class
- remove MongoDB module load from the Perlmutter config
- remove mpi4py from doc requirements
- save sub-agent config into Registry
- sub-agents are no daemons anymore
- update documentation for Polaris (GPUs assignment)
- update launcher for `Agent_N`
- update sub-agent config (in sync with the agent default config)
- update "Describing Tasks" tutorial
- use RMInfo `details` for LM options


--------------------------------------------------------------------------------