Twittersphere

Adds a new column that enables working out which of several versions of a tweet were directly collected. This helps to link a directly collected tweets and users back to the context in which they were collected, which can't be done with the directly_collected_* tables.

Querying tweet_latest and user_latest should continue to use the directly_collected_* tables, as the latest version of a tweet may or not be directly collected.

- change mentioned_user_id to be of type integer instead of string, for consistency with all other user_id fields
- allow public_metrics.listed_count to be optional with a default of 0

Adds new fields for the edit related information of tweets to the `tweet_at_time` table:

+ edits_remaining
+ is_edit_eligible
+ editable_until
+ min_edit_history_tweet_id

And adds a new table `tweet_edit_history` to allow easily lookup all versions of a tweet via the `min_edit_history_tweet_id` field.

New functionality:

- entity annotations are extracted from the JSON if present, and stored in the `tweet_annotation` table.
- Labels can now be provided on inputs to differentiate different types of collections that need to be compared in some way. As an example, this can be used on the command line to create a single database with a distinction between two different geographic regions:

shell
twittersphere prepare --label AU geo_tweets_au.json geo.db
twittersphere prepare --label US geo_tweets_us.json geo.db

The label is applied in the `collection_context` table, and can be used in sql queries along with the `context_id` column in the main user and tweet tables to draw distinctions.