Label suggestions with empty input are too slow on systems with many labels

Issue Summary

This is reproducible on Data Center: yes

This is a scope split of JRASERVER-73769.

Steps to Reproduce

1. Populate the label table with millions of entries
2. Expand the dropdown of a Label field in any Issue and observe how it takes long to run and the Thread is high on CPU load on the server

Expected Results

The autocomplete would not take so long to run everytime.

Actual Results

The thread becomes long running on the server and is usually not even useful to the end-users, as they may start typing labels to narrow the search, but the original request is still running.

Workaround

Currently, the only way to spare the server (and end-users) of these empty requests is to block them in the Proxy or Jira:

• How to block Label type fields autocomplete suggestions and other requests in Jira