Fix capacity snapshot selection for zero-capacity symbols

[aiSynergy37]

Description

Fixes an edge case in CapacityEstimate.UpdateMarketCapacity where snapshot selection could pick a symbol with no valid market-capacity data (Trades == 0 or MarketCapacityDollarVolume == 0).

This could cause a stale/delisted zero-capacity symbol to drive snapshot logic and trigger unnecessary margin model calls.

Changes

• Filter snapshot candidates to symbols with usable capacity inputs:
  • Trades > 0
  • MarketCapacityDollarVolume > 0
• Use FirstOrDefault() for candidate selection and skip scaling logic when no valid candidate exists.
• Preserve existing reset/removal housekeeping for symbol capacity entries.
• Add regression test:
  • CapacityEstimateTests.UpdateMarketCapacitySkipsSymbolsWithoutMarketCapacityData
  • Verifies we do not query reserved buying power for symbols without market-capacity data.

Validation

• dotnet build Tests/QuantConnect.Tests.csproj --no-restore -v minimal ✅
• dotnet test Tests/QuantConnect.Tests.csproj --filter FullyQualifiedName~CapacityEstimateTests.UpdateMarketCapacitySkipsSymbolsWithoutMarketCapacityData --no-build -v normal -m:1 ⚠️
  • Test host aborts in this environment due Python.NET GIL finalizer crash:
    • System.InvalidOperationException: GIL must always be released...

Closes #6523