erp5_web_mynij_search: Fix downloading index from torrent, load source now check if there is a torrent for download

When loading new source from store link, check if there is attached torrent and try to download it. If not possible to download after a timeout, register a build for the source.

If get a webpage during build return 401, 403, 404, etc ... we just skip the page and continue

Update existing document if there is a source with same title

When loading html element result, browser will fetch all images, links, css, ... present in html string. We removed all url before load html element.

Also cleanup textContent so that there is not useless content.

Remove Views and Actions for other page when navigating to Search and Activities menu in left panel

Default storage is local
Remove Storages from panel menu
Add some useful links to panel

Also fix configuration file to be json compatible.

RTC configuration can be saved in to file and will be downloaded by the app

built index many time should not add the same page again in the idexeddb. Unique Id is generated from the page URL

When build fail, it will loop trying to build again, we just has to change the state to failed and user can decide to restart build when he see the failed notification.

Torrent are compressed before seed
torrent blob is saved as attachment and can be reloaded quickly without rebuild everything
There are more details on download and seed progress

add magnet_uri to souce form
add new tracker URL
destroy torrent when download is finished (to prevent client to automatically start to seed the torrent)

Use JSzip to compress data exported to reduce size of downloaded/uploaded file

Can now import a json file containing a list of index defintion dict which is a title and links.

Also possible to export all current index definition in a json file a share to others users.