Created a generic PagedDatabaseObserver (common logic for conversation & gallery paged database queries and observation)
Updated the MediaGallery to use the PagedDatabaseObserver
Split the interaction and thread data queries for the conversationVC
Removed the old OWSProfileManager and OWSUserProfile (refactored)
Removed all the old TSInteraction/etc. types (replaced with new database types)
Removed the old Quote models (refactored)
Removed the old Attachment models (refactored)
Removed the old recipient identity (unused)
Deleted a number of other unused files
Fixed up quote attachment sending and retrieval
Validated attachment sending and retrieving is working correctly
Re-added the AttachmentUploadJob migration
Removed some more legacy code
Added back logic similar to the pre-processing de-duping logic (was resulting in "unsent" messages reappearing)
Added a number of updated view files
Refactored the SignalApp class to Swift
Fixed a horizontal alignment issue in the ConversationTitleView
Fixed an issue where expiration timer update messages weren't migrated or rendering correctly
Fixed an issue where expiring messages weren't migrated correctly
Fixed an issue where closed groups which had been left were causing migration failures (due to data incorrectly being assumed to be required)
Shifted the Legacy Attachment types into the 'SMKLegacy' namespace
Moved all of the NSCoding logic for the TSMessage
Refactored the AppDelegate from Objective C to Swift
Updated the HomeVC to use GRDB
Refactored a number of the Job types to be driven via GRDB and the new JobRunner
Fixed a bug where the LinkPreviewView wouldn't render correctly in dark mode
Added the rest of the interaction structure to the database (testing some migration logic now - still needs to be finalised)
Updated the YDBToGRDB migrations to wrap their inserts in autorelease pools (helps memory slightly, unfortunately it's caching the YDB data which uses the most memory but we have opted for speed over RAM at the moment)
Updated the MockDataGenerator so it should now "chunk" the code generation (crazy large figures were previously resulting in excessive memory usage)