Tradingstrategy

This is a release to get Trading Strategy client towards live trading.

- Created `tradingstrategy.universe.Universe` helper class to encapsulate trading universe
- Added convenience method `PairUniverse.create_from_pyarrow_table_with_filters` so creating different trading universes is less lines of code
- Added convenience method `tradingstrategy.utils.groupeduniverse.filter_for_pairs_on_exchanges`
- Added helper methods to construct single trading pair universe, see `PandasPairUniverse.create_single_pair_universe`
- Separate QSTrader to an optional dependency. Please install as `pip install tradingstrategy[qstrader]`.
This is because QSTrader depends on Seaborn that depends on SciPy that does not work so good on Apple M1 macs.
- Add the default connection and read timeout 15 seconds to all client downloads

- Reworked how QSTrader framework integration works. QSTrader was originally designed for stock markets, so it required few changes to make it more suitable for 24/7 cryptomarkets. **Warning**: QSTrader and its integration are both in beta. More work is needed to be done e.g. in fee calculations to make the integration smooth. See example on [PancakeSwap momentum trading algorithm](https://tradingstrategy.ai/docs/programming/algorithms/pancakeswap-momentum-naive.html).
- Added [portfolioanalyzer](https://tradingstrategy.ai/docs/programming/api/portfolioanalyzer.html) to make timeline of portfolio construction over the time. See example on [PancakeSwap momentum trading algorithm](https://tradingstrategy.ai/docs/programming/algorithms/pancakeswap-momentum-naive.html).
- Include `seaborn` as a dependency as it is required for `qstrader` plot output, otherwise `qstrader` was crashing.
- `PairUniverse` supports indexed lookups to deal with the high pair count.
- More convenience methods for candle and liquidity manipulation.

- Adding the preliminaty price impact analysis. [See the example notebook](https://tradingstrategy.ai/docs/programming/examples/price-impact.html). [Read the blog post about the liquidity formation](https://tradingstrategy.ai/blog/announcing-support-for-liquidity-charts).

- Make fastquant dependency optional, as it was causing installation issues with its ccxt dependency

- Updated inline interactive setup from within Jupyter notebook

- Polish README, release process and docs