Solr Relevancy Epic

[pgwillia]

Is your feature request related to a problem? Please describe.
Data passing through the Discovery architecture is a black box to most of us. There are some issues that @seanluyk has been collecting about relevancy:
https://docs.google.com/document/d/123osY1nFFQmUJX7jTVz72PcY6O5cbK1MCYa4XPhWVPo/edit

Describe the solution you'd like

• These to be better understood.
• More clearly documented.
• Improved and intuitive for users.

[pgwillia]

This is a summary of what I've found about mapping of data within discovery
https://docs.google.com/spreadsheets/d/18bNfP3gE2zdorVpGbGCNNTTrSirrRNGjtpT72zWdVhk/edit?usp=sharing

[pgwillia]

We've got titles and subtitles being indexed separately. This phrase is composed from both fields and will match neither. #1309

Our minimum should match configuration leads to some confusing edge cases. british missions south pacific (319) vs british missions "south pacific" (1,358)
#1508

[pgwillia]

This long running process is re-running logs against the new Solr configs to check for unexpected changes.

[pgwillia]

Can also parse the log for visualizations. This is the ELK stack (elastic search, logstash, kibana)

[seanluyk]

@pgwillia very cool, this is going to be really handy going forward!