Manually remove identity consolidator

[jhonabreul]

Expected Behavior

A manual call to SubscriptionManager.RemoveConsolidator(symbol, consolidator) removes registered consolidator when called from Python algorithm

Actual Behavior

When the consolidator is resolved to an IdentityDataConsolidator (might also be in other cases) SubscriptionManager.RemoveConsolidator doesn't remove the consolidator.

Potential Solution

The source of the bug might be because RegisterIndicator will try to convert the consolidator PyObject to an IDataConsolidator and fails, wrapping the object in a DataConsolidatorPythonWrapper. So after this, the actual stored consolidator referenced is wrapped and SubscriptionManager.RemoveConsolidator won't find it.

We might need to detect the wrapper and remove by its underlying python model.

Reproducing the Problem

Attached backtest: https://www.quantconnect.com/terminal/clone/-/ba9a391ee0a14a56cdc0729838675305/clone-of%3A-calculating-red-seahorse

class TestAlgorithm(QCAlgorithm):

    def initialize(self):
        self.set_start_date(2013, 10, 7)
        self.set_end_date(2013, 10, 11)

        self.spy = self.add_equity("SPY").symbol

        self.consolidator = self.resolve_consolidator(self.spy, Resolution.MINUTE)

        name = self.create_indicator_name(self.spy, "close", Resolution.MINUTE)
        identity = Identity(name)
        self.indicator = self.register_indicator(self.spy, identity, self.consolidator)

        self.schedule.on(self.date_rules.today, self.time_rules.before_market_close(self.spy), self.remove_consolidator)

    def remove_consolidator(self):
        self.subscription_manager.remove_consolidator(self.spy, self.consolidator)

        consolidator_count = sum(s.consolidators.count for s in self.subscription_manager.subscriptions)
        if consolidator_count > 0:
            raise Exception(f"The number of consolidator should be zero. Actual: {consolidator_count}")

NOTE: This fix is required in order to be able to use PearsonCorrelationPairsTradingAlphaModel in PearsonCorrelationPairsTradingAlphaModelFrameworkAlgorithm and any other algorithm. As of now, that algorithm is not importing the actual python model (with from Alphas.PearsonCorrelationPairsTradingAlphaModel import PearsonCorrelationPairsTradingAlphaModel) so it's using the C# one.
The Python PearsonCorrelationPairsTradingAlphaModel does a manual consolidator removal that has this bug and the algorithm final assertion for the consolidators count fails when using the python model.

System Information

Windows 11
QC Cloud

Checklist

• I have completely filled out this template
• I have confirmed that this issue exists on the current master branch
• I have confirmed that this is not a duplicate issue by searching issues

• I have provided detailed steps to reproduce the issue