Betfair: Amend order qty support + a few minor fixes

[imemo88]

Pull Request

Apologies for a somewhat large PR. I tried to keep it small.

Feature:

• The Nautilus modifyOrder API supports price and quantity amendments. Betfair cannot modify both in one atomic transaction. The current BF adapter only support price amendments but this change will allow the user to modify price OR quantity using the modifyOrder API.

Fixes:

• ModifyOrder (price) failures did not correctly parse failures (ExecutionReportStatus.FAILURE vs InstructionReportStatus.FALURE)
• Multiple execution.connect() calls will spawn multiple update_account_state() coros. Moved create_task(update_account_state) out of the connect method AND converted it to a do-while method as the sleep should be called first in a while True-loop in case of exceptions.
• The rust logger? does not support ", exc_info=True)" in log messages so opted to use traceback.format_exc instead.
• SUBSCRIPTION_LIMIT_EXCEEDED: Treat it similar to MAX_CONNECTION_LIMIT_EXCEEDED as SUBSCRIPTION_LIMIT_EXCEEDED does not close the connection. The connection will stay broken presently.

Type of change

• [x] Bug fix (non-breaking change which fixes an issue)
• [x] New feature (non-breaking change which adds functionality)

How has this change been tested?

In prod, but I suspect I may need to submit a few modifyOrder fixes once I get more runtime

[limx0]

LGTM - Thanks @imemo88

@cjdsellers this is good to go from my end

[limx0]

This following is causing an infinite loop in test_account_statement

class BetfairExecutionClient(LiveExecutionClient):
    def __init__() -> None:
        .... # omitted 
        self._init_scheduled_tasks()    <---- causing infinite loop

    def _init_scheduled_tasks(self) -> None:
        self.create_task(self.account_state_updates())

Multiple execution.connect() calls will spawn multiple update_account_state() coros.

I think we're still going to want to start this in the connect method. If we're creating multiple instances we either need to skip if one exists already or better yet clean it up before we call connect again.

A better solution would probably be saving a reference to the running account state task and shutting it down in the _disconnect method - which gets called on reconnection.

[imemo88]

Thanks @limx0, I'll have a look at the suggestions!

[cjdsellers]

Closed from #1751 and will also credit @imemo88 in the release notes.