[Feature]: support more custom options for agent.recordToReport()

[xxleyi]

What problem does this feature solve?

There are quite a few capabilities in the current official report that are not exposed externally through agent.recordToReport(). However, if they can be exposed, it would be very user-friendly for users who carry out secondary development based on midscene.

The main capabilities that can be exposed are as follows:

• Under the click step, midscene currently displays multiple screenshots, such as the screenshots before and after the click. Can the custom step expose configuration items to allow users to pass in multiple screenshots in a single API call?
• The type of the total step is now hardcoded as Log when calling agent.recordToReport(). Can this support user pass custom value so that users can accurately convey custom intentions?

What does the proposed API look like?

interface RecordToReportOptions {
content?: string;

// type?: string - 可选，截图的类型，如果未提供，则类型为 'log'。
type?: string;
// customScreenshotData?: Object[] - 可选，自定义截图数据数组，不传时自动截图，传入时根据 base64 编码展示截图。
customScreenshotData?: {
// description?: string - 可选，截图的描述。
description?: string;
// base64: string - 必选，截图的 base64 编码。
base64: string;
}[];
}

function recordToReport(title?: string, options?: RecordToReportOptions): Promise;