PR for 0.2.0 Release

[szwiep]

PR for 0.2.0 Release

- Support for Sentinel 2 imagery

• Effected files: enums.py, indices.py, io.raster.py
• Create Platform enum; supports landsat tm, etm, oli, and sentinel 2
• Users provide platform(s) through platform= parameter in read_and_stack_tifs
• Decorators added to index methods for setting platform compatibility and band requirements for each index

- Update to recovery target methods

• Effected files: restoration.py (ReferenceSystem), recovery_target.py
• mean replaced with median
• Historical recovery targets are now per-pixel (band, y, x), reference targets are still per-polygon (band)
• ReferenceSystem uses a historical_reference_system parameter to choose type of recovery target method to compute in ReferenceSystem.recovery_target()

- Add plotting for spectral trajectories of restoration site

• Effected files: restoration.py, timeseries.py
• Add plot_spectral_trajectory() method to RestorationArea. Method plots trajectories for each band; includes restoration, disturbances, and reference windows if relevant. Implementation is fairly slow, likely due to Pandas.

- Support for band naming

• Effected files: io.raster.py
• Add band_names= parameter in read_and_stack_tifs to allow users to map band numbers to band names. Partially addresses Allow band numbers, not just common names #22 (feedback from VTT)

- Support for additional bands

• Effected files: enums.py
• Support for Red Edge and Aerosol Coastal band, partially addresses read_and_stack_tifs throws error for valid but unsupported band common names #23 (feedback from VTT)

- NumPy computation

• Effected files: io.raster.py
• Added array_type= parameter, to allow users to choose NumPy or Dask arrays; if NumPy is chosen, all computation run in spectral-recovery will be NumPy based (locally). Useful for comptuation over smaller datasets.

- Allow str inputs from users

• Effected files: timeseries.py, io.raster.py, indices.py
• Users now pass str inputs for index choices in compute_indices and platform in read_and_stack_tifs
• restoration start, disturbance state, and reference years for RestorationArea(...) can now be passed as strs instead of Timestamps

- Loosen spatial check

• Effected files: timeseries.py
• Allow close-to-zero distance in contains_spatial when checking whether images contain restoration/reference polygons. Accounts for rounding errors that pop up between almost-equal bounding boxes

- Update docs

• Effective files: docs\overview_API.ipynb, docs\test_data\
• Test data updated to Landsat timeseries used in validation process, restoration polygon updated to 2005 wildfire. Timeseries clipped to polygon to make dataset smaller.
• overview_API.ipynb updated to reflect new changes in this release.
• Add example of how to write rasters with rioxarray to end of overview_API.ipynb, closes Add documentation for writing TIFS #24 (feedback from VTT)

- General clean up

• Improve docstrings for public methods
• Add more corner cases to unit tests
• Lint codebase using pylint, see Lint codebase #36 for descriptions

[KavlinClein]

It's looking good. A couple of small things to change for the next update, but nothing major from what I could see!