make deployment dense

[maxheld83]

this may contribute to slowing down on large requests as per #133

[maxheld83]

aka per-app scaling

[maxheld83]

upon further inspection, this is unlikely to bring a lot of value now, until we have async and some perf/load metrics.
This would allow us to ensure that instances hosting metacheck don't host any of our other apps at the same time.
But there'd still only be one metacheck instance, requiring heavy async work.

This would also only apply once there are more than 1 instances, which is unlikely to be the case for most hours of the year.

And given that even with heavy async, shiny always is a bit of a bottleneck, we might actually be better off having more "sparse" instances, rather than fewer, denser instances, because in the former case, we'll have more shiny endpoints to talk to.

Details in: https://docs.microsoft.com/en-us/azure/app-service/manage-scale-per-app

Moving out of current milestone.