Move AUDIT_DATASET to coreCIF

[vaitkus]

The AUDIT_DATASET category seems to be applicable outside of the multiblock approach used for the representation of complex data. For example, if I understand the example correctly, this approach could be used to link the calibration data block to the measurements of several regular crystals that follow it.

As such, it might make sense to move it to the coreCIF dictionary that is imported by MultiBlock dictionary either way.

[jamesrhester]

The thinking behind putting the definition in the multiblock dictionary is that, in order for the contents of multiple data blocks to be interpreted as parts of a larger whole, there must be at least one foreign key link between tables in separate data blocks. Otherwise, each data block contains a disjoint group of tables that have no formal relationship to each other and therefore can't constitute parts of a dataset. I'd go further and say, once we've completed the multi-block work, every single category must have at least one in-going or out-going foreign key link.

The multiblock dictionary creates (almost?) all such links, so it makes sense for audit_dataset to be there.

[vaitkus]

My original point stands corrected. However, I still see the utility of having this category in coreCIF.

Historically, most of the produced CIF files only had a single data block and I image that this trend will continue for the foreseeable future. It would be nice to be able to assign these CIF a dataset ID without also having to import the more complex data model provided by multiblockCIF.

If this does not seem relevant, feel free to close the issue.

[rowlesmr]

In the end, I foresee multiblock being subsumed into core; multiblock is an experiment showing how it works without mucking around with core. Once everything is set, it should just transfer straight across.

[jamesrhester]

Historically, most of the produced CIF files only had a single data block and I image that this trend will continue for the foreseeable future. It would be nice to be able to assign these CIF a dataset ID without also having to import the more complex data model provided by multiblockCIF.

This is a good point. I have no objections to transferring it to core.

[vaitkus]

Resolved by merging #62.