Changes
* Adapted the new `Group` system implemented in `aiida_core >=1.2.0` for pseudopotential families, the change is backward incompatible.
* Unified families for `UspData` and `OTFGData` to a single `OTFGGroup`. Two types of pseudopotentials can be mixed up in a single group. Mixing `UpfData` is not supported through this API yet, but can be added manually.
* Added command `verdi data castep-otfg migrate` to migrate the old families. This command **MUST BE RUN ONCE** after the upgrade, otherwise pseudopotential family matching will **NOT WORK**. NOTE: old groups will not be deleted using this command for now.
* Allow move the old task specific calculations inside `CastepCalculation` - it can now accept additional kpoints list/mesh for specific tasks. Entry points are removed for the task specific calculation as well. *Note: all kpoint meshes defined will be used as Monkhorst-Pack grids by CASTEP, which are not always Gamma-centred.*
* CASTEP will skip geometry optimisation if it found there is nothing to optimise. Now the plugin will detect this and add the initial structure as the output as the structure is indeed optimised in full.
* added `bypass` option in `relax_mode` for `CastepRelaxWorkChain` for skipping any higher level control. This is primarily to make the project work with `aiida-common-workflows`.
* Make `kpoints_spacing` input for `CastepBaseWorkChain` respect the `pbc` information of the input structure. Previously it was assumed that all input structures are periodic. Now this information will be used for setting the kpoint grid. Directions that are non-periodic will not have the span along it. This means that a Gamma-point only calculation will be performed, if all three directions are non-periodic and the spacing input is ignored.
Bugfix
* No longer using `Node` instant as default for workchains.
* Fix a bug where the forces are not resorted. CASTEP will internally sort the atoms, the plugin reorders the atoms such that the output structure have the same atoms order as the input structure. However, the same is not applied to the forces and velocities previously. Calculations where the input structure not having species grouped and sorted in the ascending atomic number had forces in the wrong order.