Align approaches to mixing for symptom survey aggregates

[nmdefries]

The old pipeline for community responses did not apply mixing. To reproduce it, we currently ignore the mixed weights for some aggregates. Once a better mixing/weighting scheme is chosen, all signals should use it. Once a new approach is chosen, we may need to reissue old signals or update public-facing documentation about the change.

[capnrefsmmat]

I believe we can shift to mixing all signals. I'm told that trimming weights is standard practice, and we can adopt this for consistency.