[Feature]: Support business-context-aware end-to-end test generation and deterministic workflow compilation

What problem does this feature solve?

1. Support injecting requirements documents, business rules, field enumerations, API definitions, and page descriptions into the Agent.

2. Support breaking down business scenarios into structured steps instead of relying on a single long aiAct instruction.

3. Support control-flow capabilities such as if/else conditions, loops, retries, and step dependencies.

4. Support common admin-platform scenarios such as combined filtering, table traversal, and pagination.

5. Support converting a successful AI-driven execution into a reusable Playwright or YAML workflow.

6. Prioritize stable selectors and use visual element localization as a fallback.

7. Support UI assertions, API assertions, and structured data assertions.

8. Record why each step was skipped, failed, or terminated early.

What does the proposed API look like?

Currently, midscene can not combine complex business logic process with ui automatic operation，so we often encounter issues like midscene can not fully complete a full process like below
Step1: open a page
Step2: click the tab
Step3: find the filter button
Step4: if has Succsss Option,click it,if not, go to another page and do some config...
like abover complex test cases，midscene can't do this fully tests ,if only depends frontend project codes,Anyone has suggestions will be appreciated，Thanks in advance~

No files, tests, or entry points are named. Start by searching the repository for the Agent, aiAct, and existing Playwright or YAML workflow generation paths, then map the requested business context, control flow, assertions, selectors, and execution-status reporting to current APIs. Done would require an agreed, testable scope for these capabilities rather than the broad feature list alone.