Improve publishing formats and throughput on common endpoints

[jdpye]

Making general improvements to our publishing strategy into commonly used endpoints (GeoServer, ERDDAP) to mitigate some of the troubles we've experienced in publishing and hosting full-resolution data to those spaces.

Historically we've been able to tip over GeoServers by requesting a few million rows, so for detection data, this might mean an ERDDAP dataset that is lean but informative and can refer to external datasets easily.

[jdpye]

Was mentioned by @aubrivliz that exfiltrating data from GeoServer in XML format was the least processing-intensive to GeoServer itself and may allow for some performance over high rowcount querying. We will test this out a bit on our larger table layers and find out whether this solves our bottleneck.

[jdpye]

Steps ETN took:

updated Geoserver (v 2.?? ?)
optimization.
updated PostgreSQL (14.x now)
indexing of the Geoserver-mapped tables shouldn't have changed but will see.

TODO: OTN to try again and see if issue persists.
OTN to beg ETN for optimization tips for the reverse flow.