early out for non distributed arrays

[mtar]

Due Diligence

• General:
  • title of the PR is suitable to appear in the Release Notes
• Implementation:
  • unit tests: all split configurations tested
  • unit tests: multiple dtypes tested
  • NEW unit tests: MPS tested (1 MPI process, 1 GPU)
  • benchmarks: created for new functionality
  • benchmarks: performance improved or maintained
  • documentation updated where needed

Description

The collective MPI communication functions check if the communicator size is one and copies the send buffer to the receive buffer if available. Additionally, this PR removes direct access to mpi variables outside of the communication module.

Issue/s resolved: #2326

Changes proposed:

• access MPI variables through own communicator, not directly
• make general communicator abstract
• add early out in collective mpi communication functions
• use newer type hints

Type of change

Refactor

Memory requirements

Performance

Does this change modify the behaviour of other functions? If so, which?

yes / no