Optimize `write_records` method for speed and to ensure they are saved onto DB

[eddowh]

Highly recommended to bump version after merging.

[eddowh]

I was thinking - when we want to write a list of records into the DB, why don't we check the list for duplicates and remove them before committing, like such:

1. We're given a list of tuples [(datetime, uuid, key, value) , ... ] corresponding to records.
2. Remove the duplicates in the list of tuples by matching against datetime, uuid, and key (which are all primary keys when we define the Record schema in SQLAchemy)
3. Map through the list of records tuples into a list of Record objects [Record<ts, uuid, key, value> , ...]

4. By now, the list of Record objects can be added and committed to the database, which is simply done with:

   session.add_all(list_of_Record_objects)
   session.commit()

This way, we don't have to atomize the list of records when trying to commit them. It'd be faster and I suppose the code will also be more readable.

[hrbonz]

You can't know if duplicates are all in the data about to be pushed so let the db deal with that and handle the situation.

[eddowh]

Method implementation is outdated due to inconsistency with data format of input. See #30.

Closing.