Investigate collection fusion methods for vector search outputs

What is the best way to merge vector search results from the bills and the ballot questions collections? Is there a way to calibrate a distance metric that is applicable to both? The goal is to be able to compare vector distnace results between the two types of content apples to apples.

_Originally posted by @nesanders in [#2218](https://github.com/codeforboston/maple/pull/2218/changes/eaea9c20baf41cd8af240f7d734b6623aae16564..c0eaac145c59de297f8867742452f6adf3227dfc#r3858719135)_