Refactor index to better support unread messages #496
Description
Currently, very few unread messages are retrieved during initial load. Due to lack of data about unread messages, issues like #434 cannot be solved. Solution is to:
Use the unread message data from model.initial_data['unread_msgs'] from server. Add data structures to the index to store info about unread messages, and to prevent retrieval being costly. When marking message as read, remove the message from the newly added data structure
For more info, refer to https://chat.zulip.org/#narrow/stream/206-zulip-terminal/topic/.5BCode.20help.5D.20how.20does.20index.20work.3F