might be interesting to consider and document the difference in meaning vs MissingResourceSpecification, also in mpi_prefix_composer (and maybe merge?)

/There are three places where the executors raise the same or similar error.

1. when passing the resource_specification parameter to the htex, it raises the InvalidResourceSpecification from validate_resource_spec method. https://github.com/Parsl/parsl/blob/master/parsl/executors/high_throughput/executor.py#L342

2. ThreadPoolExecutor also raises the UnsupportedFeatureError when passing resource specification to it, but the raised error is in parsl.executors.error. https://github.com/Parsl/parsl/blob/master/parsl/executors/threads.py#L54

3. WorkQueueExecutor raises two errors similar to this but as a Base ExecutorError, one when there is invalid resource_specification ; https://github.com/Parsl/parsl/blob/master/parsl/executors/workqueue/executor.py#L418 and another one when the autolabel is false and resource_specification is not right, https://github.com/Parsl/parsl/blob/master/parsl/executors/workqueue/executor.py#L429

From the above,

1. As ThreadPoolExecutor raises the UnsupportedFeatureError from the parsl.executors.error, ideally htex should also raise the same instead of raising InvalidResourceSpecification.

2. If we look at 3, the reason for the second error to be raised is the expected fields in resource_specification are missing. Here, the behaviour is similar to both the MissingResourceSpecification and InvalidResouceSpecification.

What we can do,

1. We can remove the MissingResourceSpecification and InvalidResourceSpecification from mpi_prefix_compose and merge them as ResourceSpecificationError in the parsl.executors.error. As MissingResourceSpecification is raised only when an empty resource_specification is supplied.

2. htex and threads can raise the UnsupportedFeatureError as that is more reasonable than ResourceSpecificationError.

3. WorkQueueExecutor and mpi_prefix_compose can raise the ResourceSpecificationError with appropriate Messages.