This implements a huge change, where Fresh is removed as a framework and serving files, allowing more control over importing, bundling, and serving files and components.
The biggest challenge was to continue making sure that there weren't too many places to look into for import versions, and `PasswordlessPasskeyLogin.tsx` became a prototype in migrating a component to fully SSR, no need for frontend parsing (via Babel) or bundling (via a custom-script, downloading frontend dependencies from esm.sh). Still, there are too many components to migrate like that, and it's all working, so I likely won't even attempt it unless there's some bug, new feature, or security vulnerability to address that warrants a rewrite of those.
This also updates all dependencies (except `@libs/xml` because that still causes some breaking in DAV endpoints), including Deno!
All other advantages can be seen in the related issues, and the breaking change this (v4.0.0) introduces is related simply to `config.email.tlsMode` (which had a deprecation warning throughout v3), and because, while I tested many things exhaustively, it's not impossible something broke that I didn't see.
Closes#141Closes#132
* Implement a more robust Config
This moves the configuration variables from the `.env` file to a new `bewcloud.config.ts` file. Note that DB connection and secrets are still in the `.env` file.
This will allow for more reliable and easier personalized configurations, and was a requirement to start working on adding SSO (#13).
For now, `.env`-based config will still be allowed and respected (overriden by `bewcloud.config.ts`), but in the future I'll probably remove it (some major upgrade).
* Update deploy script to also copy the new config file
This refactors the data handlers into a more standard/understood model-like architecture, to prepare for a new, more robust config system.
It also fixes a problem with creating new Notes and uploading new Photos via the web interface (related to #58).
Finally, it speeds up docker builds by sending in less files, which aren't necessary or will be built anyway.
This is all in preparation to allow building #13 more robustly.
