Quant-invest-lab

Added price backtest in addition to the OHLC backtest.

From Plotly to bokeh
a lot of new metrics for risk and performances
code refractoring
removed some LRU cache (caused issues...)

- Fixed bug in data provider file

- Added constants and types modules
- Return is calculated using the next Open and not the current close: more realistic
- Added returns to candle generation
- Added signal package: denoise, detrend, spectrum analysis...
- Using itertuples in backtests
- Timestamp could be natively dropped from downloaded data

- Refractored backtest the code
- Added LRU caching
- Added Expectancy metric
- Added buy-and-hold comparaison for VaR, Skew...
- Moved `reduce_dimenstionality` in `/utils.py`