* set enabled sources via settings

* remove header

* check responses

* unify edit modal, support adding to other osurce collections

* open edit modal from view

* add community to scope

* fetch image for margin/community

* renames and don't refetch failed images

* enable multiple bookmark sources

* cleanup imports and return collections/tags from datasource interface

* render unified filters

* simplify/consolidate filters, multi filter select

* track item collecitons in view

* render collection/count on card

* refactor render into view

* fix all filter

* refactor card detial rendering

* add edit button to detial modal

* add icons

* add icons

* better styling of filters

* fix tag filtering

* unify create collection modal

* add source icon to filter menus

* add icon to chips, fix bookmark colleciton filtering

* cleanup

* bump version

* ignore invalid site.standard.documents instead of erroring out

* bump version

* desktop only until oauth sorted out

* add refresh btn

* refresh on card/collection/tag changes

* bump version

This reverts commit 1418f94d77704dfae4594a680804fa400e87d987.

* add metadata

* test pages action

* initial oauth flow

* update scope

* login in settings

* remove session from plugin class

* remove oauth modal

* add session store / refactor

handle missing block text

* render before done fetching feed

* initila work for redirect to obsidian uri scheme

* switch back to browser client

* cleanup storing session/agent

* cleanup

* better html page

* better settings layout

* remove lex generated code

* bump tag

* publishing: Pull title from H1

Allows the title to be sourced from the published document's first H1
header. The order of precedence for determining the document's title is
now:

- `title` frontmatter field
- First present H1 header
- Filename

* refactor header extracting

* enable via settings

* fix lint

---------

Co-authored-by: Isaac Corbrey <isaac@isaaccorbrey.com>

The `path` frontmatter field in notes being published is now passed on
to the resultant `site.standard.document` record, and the `url` field
written post-publish now uses that path and is sanitized to prevent
double slashes.

Producing a Pckt.blog artifact now includes the correct facets for
encoding headers, links, and styling.

* handle existing clipped document

* bump version

* resolve actors with slingshot

* progressively render publiactions

* bump version

* rename plugin, update readme

* add feed command

* view list of publicaiton's docs

* clip documents to vault

* add frontmatter

* use bsky link for author

* trim whitespace entities

* refactor client to cache responses

* wrap client

* initial work for standard.site/leaflet

* update document

* set frontmatter for published doc

* initial work for pckt

* remove publicaitons view

* lint and type fixes

* add agent resolver

* add view to list subscribed publications

* bump version

add support for margin.at