Allow multiple DOID terms for single evidence items

[obigriffith]

Occasionally an evidence item (e.g., describing a clinical study) determines a clinical relevance for a variant to multiple cancer subtypes. While these can be bucketed under some generic parent term (e.g., 'cancer') this loses some potential important detail. Another solution is to break into multiple evidence items but this creates a large curation burden. It has been requested to allow entry of multiple disease (DOID) terms for the single evidence item.

[arpaddanos]

Another angle on this is the generation of more complex disease entities such as Melanoma Brain Metastasis or treatment contexts such as Neoadjuvant Breast Cancer.

[arpaddanos]

Seems related to #893