Translation memory and search backends

[nijel]

Describe the problem

The current trigram-based lookup for translation memory and full text search is not performing well in some situations. We have several hacks to make it work better like skipping trigram index when it would not perform well. Let's look if there are some viable alternative approaches.

Solution brainstorm

• Search using an external solution
  • ElasticSearch / OpenSearch
• PostgreSQL plugins/extensions
  • https://github.com/timescale/pg_textsearch
  • https://www.paradedb.com/

Describe alternatives you have considered

No response

Screenshots

No response

Additional context

• Search should ignore hotkey symbols #1861