Allow similarity measures to be used when there are multiple upstream gauges in SCBC

[arbennett]

In the docs we say:

When a site has multiple upstream gauge sites as tributaries we simply choose the closest.

This likely is a source of errors with situations where a gauged tributary is closer to a downstream gauge than the nearest gauge on the main stem. We should allow this to be selected just like when there are no upstream gauges.