Support custom column selection in OHLC queries and extend CCXT with configurable exchange-specific fields

- Add `columns` parameter to `get_ohlc_async` and pass through to Iceberg queries
- Replace hardcoded Binance field extraction with declarative `EXCHANGE_OHLCV_EXTENSIONS` config
- Add `applyScale` helper for field-specific transformations (ms_to_ns, price, size, int)
- Support `complementOf` spec for derived fields (e.g., sell_vol from total - buy_vol)
- Apply extensions dynamically in `convertToOHLC` and gap-filling logic
- Remove redundant column filtering in DataAPI (now handled upstream)

sandbox/dexorder/impl/data_api_impl.py

@@ -166,13 +166,11 @@ class DataAPIImpl(DataAPI):
             period_seconds=period_seconds,
             start_time=start_nanos,
             end_time=end_nanos,
-            request_timeout=self.request_timeout
+            columns=columns_to_fetch,
+            request_timeout=self.request_timeout,
         )
 
-        # Select only requested columns (filter out metadata and unrequested fields)
         if not df.empty:
-            available_cols = [col for col in columns_to_fetch if col in df.columns]
-            df = df[available_cols]
             self._bars_fetched += len(df)
 
         return df

sandbox/dexorder/ohlc_client.py

@@ -5,7 +5,7 @@ OHLCClient - High-level API for fetching OHLC data with smart caching
 import asyncio
 import pandas as pd
 import logging
-from typing import Optional
+from typing import List, Optional
 from .iceberg_client import IcebergClient
 from .history_client import HistoryClient
 from .symbol_metadata_client import SymbolMetadataClient
@@ -90,6 +90,7 @@ class OHLCClient:
         period_seconds: int,
         start_time: int,
         end_time: int,
+        columns: Optional[List[str]] = None,
         request_timeout: float = 120.0
     ) -> pd.DataFrame:
         """
@@ -123,7 +124,7 @@ class OHLCClient:
         end_time   = ((end_time   + period_nanos - 1) // period_nanos) * period_nanos  # exclusive
 
         # Step 1: Check Iceberg for existing data (run in thread — scan.to_pandas() blocks ~3-5s)
-        df = await asyncio.to_thread(self.iceberg.query_ohlc, ticker, period_seconds, start_time, end_time)
+        df = await asyncio.to_thread(self.iceberg.query_ohlc, ticker, period_seconds, start_time, end_time, columns)
 
         # Step 2: Identify missing ranges — pass df to avoid a redundant Iceberg scan
         missing_ranges = self.iceberg.find_missing_ranges(
@@ -149,7 +150,7 @@ class OHLCClient:
             raise ValueError(f"Historical data request failed: {result['error_message']}")
 
         # Step 5: Query Iceberg again for complete dataset (run in thread)
-        df = await asyncio.to_thread(self.iceberg.query_ohlc, ticker, period_seconds, start_time, end_time)
+        df = await asyncio.to_thread(self.iceberg.query_ohlc, ticker, period_seconds, start_time, end_time, columns)
 
         return self._apply_decimal_correction(ticker, df)