Refine imports for type hints in Bybit

[sunlei]

Pull Request

Refine imports for type hints in Bybit.

Type of change

Delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

[codspeed-hq]

CodSpeed Performance Report

Merging #2149 will improve performances by 26.3%

_{Comparing sunlei/bybit-refine-imports (0514414) with develop (7392f06)}

Summary

⚡ 1 improvements
✅ 51 untouched benchmarks

Benchmarks breakdown

	Benchmark	develop	sunlei/bybit-refine-imports	Change
⚡	test_order_denied_to_dict	42.3 µs	33.5 µs	+26.3%

[cjdsellers]

Are we using TYPE_CHECKING for performance reasons here?
I'd be interested to know the improvements if you have any before-after figures.

[sunlei]

@cjdsellers

I haven’t conducted performance tests for this specific scenario, but theoretically, it should be beneficial:

• Isolate type-checking code.
• Avoid unnecessary runtime imports.
• Reduce the likelihood of circular references.

[cjdsellers]

Sure, happy to merge - but thoughts are:

• We haven't experienced any circular import issues for Bybit, that I know of.
• Departs from the standard import patterns in most other Python modules.
• Could be considered boilerplate unless a strong reason for it?

A reason in favor is the potential runtime performance improvement, would be interesting to know because it could inform heading this way for other modules too.

[sunlei]

@cjdsellers

Thank you!

• Bybit previously encountered a circular reference issue in one of its PRs, which I resolved using TYPE_CHECKING.
• I randomly checked two projects, PyTorch and Flask, and found that TYPE_CHECKING is used quite extensively.
• In my own projects, TYPE_CHECKING often serves as boilerplate code.

For further reference:

• https://mypy.readthedocs.io/en/stable/runtime_troubles.html