[feat] sekotalk v2.5 通过命令行控制每段生成长度

[wangshankun]

No description provided.

[gemini-code-assist]

Code Review

This pull request introduces a target_video_length parameter to allow more granular control over generated clip lengths. Changes include adding a CLI argument, updating the S2VInputInfo and RS2VInputInfo structures, and modifying the wan_audio_runner to prioritize this value during audio segmentation. Feedback was provided to refactor the logic for retrieving the video length from input info to make it more concise and Pythonic.

[gemini-code-assist]

为了提高代码的简洁性和可读性，可以对这部分逻辑进行重构。目前的嵌套 if 语句有些冗长，可以将其简化为更紧凑、更符合 Python 风格的形式。

Suggested change

	target_video_length = self.config.get("target_video_length", 81)
	ii = getattr(self, "input_info", None)
	if ii is not None and hasattr(ii, "target_video_length"):
	tvl = ii.target_video_length
	if tvl is not None and tvl is not UNSET and tvl > 0:
	target_video_length = tvl
	target_video_length = self.config.get("target_video_length", 81)
	ii = getattr(self, "input_info", None)
	if ii is not None:
	tvl = getattr(ii, "target_video_length", None)
	if tvl not in (None, UNSET) and tvl > 0:
	target_video_length = tvl