Trading-strategy

- Fix: Missing `chains` submodule update from the previous release

- Add: `Client.fetch_token_metadata(query_type=OHLCVCandleType.tvl_v2)` option to load /WETH quoted TVL data for Uniswap v3
- Add: `utils.token_extra_data` to load additional trading pair data
- Add: `load_extra_metadata` to load buy and sell tax for given trading pairs
- Add: `CandleUniverse(autoheal_pair_limit)` and reworked how to mitigate MEV issues on pricing data
- Add: `heal_anomalies` to get rid of MEV anomalies, second attempt - handles anomalies detection and smoothing out better
- Add: `GroupedCandleUniverse.calculate_returns` to easily get returns of each trading pair
- Add: `Client.fetch_tvl_by_pair_ids(query_type=OHLCVCandleType.tvl_v2)` option to load /WETH quoted TVL data for Uniswap v3
- Change: Use `orjson` to faster serialisation of some data

- Add: `wrangle.normalise_volume` to deal with different volume formats of Uniswap v2 and v3
- Add: Support for Coingecko data loading and metadata cross referencing. See `tradingstrategy.alternative_data.coingecko`.
- Add: `tradingstrategy.alternative_data.coingecko.categorise_pairs()` to tag trading pair data with their CoinGecko category
- Add: `deduplicate_pairs_by_volume()` to make it easier to construct trading pair baskets from open-ended universes
- Add: `Client.fetch_top_pairs(TopPairMethod.by_token_addresses)` - query the best trading pairs and token tax for given token addresses

- Fix: Allow to run without Jupyter notebook/IPython installed. Make sure you use `Client.create_live_client()` instead of `Client.create_jupyter_client()`
- Fix: Python version pindown `<3.13` instead of `<=3.12` as the pip did not allow minor versions like `3.12.7`

- Fix: Bad import line in `tradingstrategty.utils.wrangle` making the library not to import properly

- Add: `Client.fetch_top_pairs()` - create a helper function to create always expanding trading universe for external signal providers
- Add: `forward_fill(forward_fill_until)`. The default behavior is to forward fill gaps between first and last candle. However the last candle might not be updated if we load live sparse data and there has been no trades (no candles). Force the forward fill to go until a certain timestamp.
- Fix: Remove `eth_defi` imports, an optional dependency, in the core library