Evidence string generation for 2024.12 release

[apriltuesday]

Deadline for submission: 23 October

Refer to documentation for full description of steps.

Blocked by #434 and #445

[apriltuesday]

There are 460 invalid evidence strings which on manual inspection all seem to use the new clinical significance values (e.g. "likely oncogenic" or "tier ii - potential"). Along with 46 records invalid for having multiple clinical classifications in the new XML structure, this is a marked increase in somatic or oncogenic records we are dropping by not supporting the new XML structure for clinical classification.

I've extracted these from the output in this batch (invalid_evidence.json and multiple_classification.txt in logs directory) and will create an issue to look into it.