Prefix FILEID for pointer file METS creation

[ross-spencer]

This commit ensures that IDs and FILEIDs are prefixed when
generated for pointer files belonging to an AIP. The prefix along
with name clean-up procedures in the transfer/ingest workflow ensures
that the ID/FILEID remains consistent with an NC string type as
defined by the XML standard.

Connected to archivematica/Issues#660

The output will be a fileSec and structMap in a pointer file that looks as follows:

  <mets:fileSec>
    <mets:fileGrp USE="Archival Information Package">
      <mets:file ID="file-my-transfer-5bc5ac2a-951f-4bc5-9039-08ec619eef1f" GROUPID="Group-file-5bc5ac2a-951f-4bc5-9039-08ec619eef1f" ADMID="amdSec_2">
        <mets:FLocat xlink:href="/var/archivematica/sharedDirectory/www/AIPsStore/5bc5/ac2a/951f/4bc5/9039/08ec/619e/ef1f/my-transfer-5bc5ac2a-951f-4bc5-9039-08ec619eef1f.7z" LOCTYPE="OTHER" OTHERLOCTYPE="SYSTEM"/>
        <mets:transformFile TRANSFORMTYPE="decompression" TRANSFORMORDER="1" TRANSFORMALGORITHM="bzip2"/>
      </mets:file>
    </mets:fileGrp>
  </mets:fileSec>
  <mets:structMap ID="structMap_1" LABEL="Archivematica default" TYPE="physical">
    <mets:div TYPE="Archival Information Package" LABEL="my-transfer-5bc5ac2a-951f-4bc5-9039-08ec619eef1f.7z">
      <mets:fptr FILEID="file-my-transfer-5bc5ac2a-951f-4bc5-9039-08ec619eef1f"/>
    </mets:div>
  </mets:structMap>
  <mets:structMap ID="structMap_2" LABEL="Normative Directory Structure" TYPE="logical">
    <mets:div TYPE="Archival Information Package" LABEL="my-transfer-5bc5ac2a-951f-4bc5-9039-08ec619eef1f.7z"/>
  </mets:structMap>

Note the ID and FILEID in the respective instances are both prefixed with file- per the original issue.

And this structure will persist following reingest as well which is why this PR: requires artefactual-labs/mets-reader-writer#72

NB. This PR is complicated by the need to use a different metsrw. Once the code has been reviewed and approved, I will push the new metsrw and then update this PR to change the requirements so they reference metsrw-0.3.10.

[ross-spencer]

(This deleted itself running the requirements Makefile)

[cole]

Argh, I was wondering when this was added. It looks like...ipython requires appnope on os x. See here: https://github.com/ipython/ipython/blob/master/setup.py#L204

[jraddaoui]

Thanks @ross-spencer!

This looks good, but I wonder about the addition of setuptools in the test requirements ...

[jraddaoui]

Was this auto-added too? Should we remove it from the requirements file and require it somewhere else if it's really needed?

[ross-spencer]

Discussed as a team in Slack. There's a little bit of chatter here too: jazzband/pip-tools#522. The follow up is to log a new issue to refine our use of pip-compile some more to catch things like this that may be undesirable.