SPIKE: Create a script for libraries to refresh Instances against an updated MARC-Instance map

Purpose: To provide a "refresh" script that a library can run on-demand whenever their MARC-to-Instance map has been updated. See more description here: https://wiki.folio.org/display/FOLIOtips/Notes+on+Customizing+Default+MARC-Inventory+Maps+and+Release+Updates

As a staff person with responsibility for the library's SRS MARC Bib-to-Inventory Instance map
I want to be able to update existing Instances based on the SRS MARC Bibs and the Library's current map
So that I can be sure that the Inventory Instances are displaying/suppressing the appropriate data from the SRS MARC record

Scenarios

1. Scenario 1
   • Given the default MARC-to-Instance mapping rules https://github.com/folio-org/mod-source-record-manager/blob/master/mod-source-record-manager-server/src/main/resources/rules/marc_bib_rules.json
   • When a library wants to ensure that Inventory Instances reflects data from the current SRS MARC Bib as described by the current MARC-to-Instance map (which may have been changed since the last time an Instance was created/updated by an SRS MARC Bib)
   • Then provide a way to refresh those Instances

May be possible to adapt Ian Walls script, which takes 1) the current version of the SRS MARC linked to an instance and 2) the current version of the tenants' MARC-to-Instance default map, and refreshes all of the source = MARC instances.

Some information from sekjal, who has created a script that does this:
That code can be found at https://gist.github.com/sekjal/b1065f0483ee18c65b79d3d6560996ed. It runs in OpenFaas, but I'm sure it could be adapted to run as a regular ol' Python script. This function does one record at a time, so I have a separate parser take a list of UUIDs and feed it to the OpenFaas engine, asynchronously, max parallelism of 3 (any more, and SRS seems to fall over). it also assumes that a set of secrets is mounted to a specific directory, but I don't suppose that'd be a problem if it was integrated more closely into Data Import. The real problem is that it's one-at-a-time processing, so it's WICKED slow. Basically relying on SRS/SRM to do the mapping work

Link to the meeting where Ian demonstrated it: https://vod.video.cornell.edu/media/Updating+Instances+Mapped+from+SRS/1_rm5wyw5t