Pull request #1818: PIPE-3089: Fix caltable filename inconsistency between serial and parallel modes
Merge in PIPE/pipeline from PIPE-3089-caltable-filenames-might-differ-between-serial-and-parallel-modes-for-some-alma-cal-tasks to main
* commit '5a560095bf4b576af51b2d4a67e4fa35279b7235': PIPE-3089: set NamingTemplate.task on MPI/Dask workers
PIPE-3089: set NamingTemplate.task on MPI/Dask workers
NamingTemplate.task is set by basetask.execute() only when
is_top_level_task() is True, which is always False on MPI servers.
This caused worker-side data product filenames to lack the task name
prefix, breaking any logic relying on filename content (e.g. the
amplitude caltable lookup in hifa_gfluxscale).
In Tier0PipelineTask.get_executable(), resolve the Pipeline task name from
task_registry (with a ParallelFoo.Task fallback) and set
NamingTemplate.task before executing the task lambda.
