Refactor rsslay to proactively retrieve feeds

[boreq]

Goals:

• Download data in the background to always be able to return data quickly.
  • Done.
• Make the project more composable e.g. separate out markdown and html processing from constructing nostr events.
  • Created clear adapter/app/domain layers. Didn't fix everything to keep the pull request small.
• Make testing easier.
  • Testing the domain layer is easy now.
• Eliminate the bug which causes the feed to not be updated under some circumstances.
  • Unclear. Maybe we need to remove caching altogether, it is not really useful right now.

What has been done:

• http handlers no longer contain core domain logic
• http handlers no longer have direct access to the database
• http handlers no longer misuse the same types for database queries and template rendering
• http handlers no longer have access to secrets
• created an adapters layer which encapsulates database access
• created an domain layer for data validation, more core logic needs to be moved there in the future
• created an application layer to glue everything together
• cleaned up duplicated calls to libraries to convert keys to the nip19 format in http handlers
• cleaned up error handling logic in http handlers
• saving feed definitions will no longer fail quietly in case of database errors
• fixed nonsensical error handling when saving feed definitions
• events are properly retrieved in a single place in the application layer and not from multiple places
• removed duplicate code