Specialsauce

Alpha version for specialsauce 0.0.3.

Completely identical to alpha version 1. I just forgot to push the updated `specialsauce/__init__.py` containing the version bump, which led to the wrong release number on PyPi.

The release can be installed from PyPi


python -m pip install --upgrade --pre specialsauce==0.0.3a2

Alpha version for specialsauce 0.0.3.

The release can be installed from PyPi


python -m pip install --upgrade --pre specialsauce==0.0.3a1


Enhancements
- data loader module `datasets`, including `load_ngp_gap()`
- Training stress formulas from TrainingPeaks
- norm (enabling Normalized Pace and Normalized Power)
- Training Stress Score (TSS)
- Acute Training Load (ATL)
- Chronic Training Load (CTL)

Half-baked ideas

I would advise against using these, or even counting on them existing in the future.

- `core` module
- `notebooks/package_usage.ipynb`: example usage
- `notebooks/hilly_course_comparison.ipynb`: explore if a hilly course could ever be faster than a flat one
- `notebooks/everesting_slope.ipynb`: identify the terrain slope that minimizes energy expenditure for an everesting attempt.
- `util`: right now, this contains half-baked moving averages that replicate TrainingPeaks' implementations.

Changes
- speed adjustment factors
- Clip grades to the range of validity of each equation
- Allow functions to work with pd.Series in addition to scalars
- Add tests for this

Fixes
- Relax strict version requirements - just set minimum.
- Remove erroneous outdated mention of GPLv3 license.

First release. Starting with the basics.