webhttp storage can get json documents by calling http GET method. for jio allDocs,
the storage will get a file called _document_list which should contain the list
of document id that can be downloaded.

replicatedopmltree storage can convert a list of opml storage (each opml can contain
one or more sub_storage) into an indexeddb. All document in opml tree will have an
entry into the indexeddb storage, the full document is added as attachment in the
indexeddb.

Add HTTPStorage and ParserStorage (RSS/OPML).

Thanks to Alain Takoudjou <alain.takoudjou@nexedi.com> for his original work on this topic.

Headers are returned as the document json.
Body is returned as the 'enclosure' attachment.

Query extend key_schema to sort_on
querystorage: add schema to use key_schema

Migrate to Dropbox API v2

Root parameter is removed, as it must be now configured on dropbox website itself.

Thanks to Sven Franck <sven.franck@nexedi.com> for his original work on this topic.

Fix getAttachment when blob is undefined

/reviewed-on nexedi/jio!74

Add facebook storage.
Keyless queries are case insensitive (like a fulltext search could be).

/reviewed-on nexedi/jio!68

/reviewed-on nexedi/jio!66

This reverts commits
215a2cca
e391b71d
ddf61098
5d6e60dd
bafcf854

Add saferepair storage
Add scenario and scenario officejs in test

Support mixed queries and add tests

/reviewed-on nexedi/jio!63

querystorage: add full text search
cryptostorage: fix confusing exemple

/reviewed-on nexedi/jio!62

ReplicateStorage: signature_hash_key is used also for attachments.
QueryStorage: support multiline text search.

/reviewed-on nexedi/jio!61

Speed up attachment replication by using a document property to check if there was a change.
The attachment content is NEVER checked, as this configuration expect the document to be modified when an attachment is created, modified or deleted.
In case of conflict, follow the document conflict resolution result.