Pyandi

Fixed
- Fix exception when there’s no existing data (bug in stale data warning)

Added
- `raw_xml` function added to dataset and activity
- Expand key-value extras in dataset metadata
- Add a filetype dataset filter
- Add a warning about stale data ([8](https://github.com/pwyf/pyandi/issues/8))
- Add location, sector, title and description schema fields

Changed
- Make it possible to download data to a custom path
- `is_valid` should just check for valid XML
- Determine filetype from metadata (instead of XML)
- Make `activity.dataset` public
- Use custom exceptions throughout

Fixed
- Fix broken `version` property
- Fix typo so metadata works again

Added
- Add functions to fetch all datasets and all activities

Changed
- Move examples to README
- Standardise the set API with an abstract class
- Compute set size more efficiently
- Make API more consistent

Changed
- Remove download.metadata() (since metadata is now included in the data dump)
- Start using python logging
- Don’t parse XML until necessary
- Where path is useless – scrap it
- Scrap custom cache location

Fixed
- Minor README updates

Added
- Initial release