Exetera

* Refactor: Created the `DataStore` class and moved `processor` api methods onto it as member
functions
* Refactor: Simplified the creation of Writers. This can now be done through `get_writer` on
a `DataStore` instance
* Fix: Writes to a hdf5 store can no longer be interrupted by interrupts, resulting in more
stable hdf5 files
* Fix: Fixed critical bug in process method that resulted in exceptions when running on fields
with a length that isn't an exact multiple of the chunksize

* Added hdf5 import and process functionality

* Feature: provision of the `split.py` script to split the dataset up into subsets of patients
and their associated assessments
* Fix: added `treatments` and `other_symptoms` to cleaned assessment file. These fields are
concatenated during the merge step using using csv-style delimiters and escapes

* Fix: `had_covid_test` was not being patched up along with `tested_covid_positive`'
* Breaking change: output fields renamed
* Fixed up `had_covid_test` is output as `had_covid_test_clean`
* Fixed up `tested_covid_positive` is output as `tested_covid_positive_clean`
* `had_covid_test` and `tested_covid_positive` contain the un-fixed-up data (although rows may
still be modified as a result of quantising assessments by day)

* Fix: `height_clean` contains weight data and `weight_clean` contains height data.
This has been the case since they were introduced in v0.1.5

* Performance: reduced memory usage
* Addition: provision of `-ps` flag for setting parsing schema