HpCalculation: bad writing of response matrices since aiida-core 2.6

[bastonero]

When parallelizing an hp.x calculation, the HpCalculation will take some parent folders in input and copy them on the remote folder under ./out/HP. Since a long time, this was performed with a small workaround that apparently does not work anymore starting from aiida-core v2.6. The issue might me related to this commit.

The following tasks should be addressed:

• Add integration suite to verify success of an actual self consistent workchain calculation (see CI: add integration test running main workflow #72).
• Add an extra pytest to verify the correct writing of files (is it even possible? is it really needed).
• Fix the problem (maybe it's not too hard, and one simply needs to add something like (with the correct order)

            calc_info.file_copy_operation_order = [
                FileCopyOperation.LOCAL,
                FileCopyOperation.REMOTE,
                FileCopyOperation.SANDBOX,
            ]

[bastonero]

Might not be a plugin problem, but a bug of aiida-core. See aiidateam/aiida-core#6767.