add llm AI EXTRACT extension feature

[ericyuanhui]

Refactor: add ClientContext overload of TaskScheduler::scheduleTaskAndWaitOrError

Summary

Adds a scheduleTaskAndWaitOrError overload that takes a main::ClientContext*
directly, so task scheduling no longer requires a processor::ExecutionContext.
The existing ExecutionContext overload now simply delegates to the new one.

Motivation / Context

Groundwork for the LLM "AI EXTRACT" extension: the extraction pipeline may need
to schedule and await tasks extending beyond the regular execution context. Callers
who only hold a ClientContext (no ExecutionContext in scope) can now use the
scheduler without building a fake context. This is the groundwork refactor; the
AI EXTRACT extension itself will land in a follow-up commit.

[ericyuanhui]

Feel free to keep the discussion going. We are sure we can reach the most ideal state. The prior implementation draws inspiration from DuckDB. LadybugDB/extensions#60

[adsharma]

Looks good. Merging. Please update the PR description. Suggested description in the comment below.

[adsharma]

Refactor: add ClientContext overload of TaskScheduler::scheduleTaskAndWaitOrError

Summary

Adds a scheduleTaskAndWaitOrError overload that takes a main::ClientContext*
directly, so task scheduling no longer requires a processor::ExecutionContext.
The existing ExecutionContext overload now simply delegates to the new one.

Motivation / Context

Groundwork for the LLM "AI EXTRACT" extension: the extraction pipeline may need
to schedule and await tasks extending beyond the regular execution context. Callers
who only hold a ClientContext (no ExecutionContext in scope) can now use the
scheduler without building a fake context. This is the groundwork refactor; the
AI EXTRACT extension itself will land in a follow-up commit.

[ericyuanhui]

Refactor: add ClientContext overload of TaskScheduler::scheduleTaskAndWaitOrError

Summary

Adds a scheduleTaskAndWaitOrError overload that takes a main::ClientContext* directly, so task scheduling no longer requires a processor::ExecutionContext. The existing ExecutionContext overload now simply delegates to the new one.

Motivation / Context

Groundwork for the LLM "AI EXTRACT" extension: the extraction pipeline may need to schedule and await tasks extending beyond the regular execution context. Callers who only hold a ClientContext (no ExecutionContext in scope) can now use the scheduler without building a fake context. This is the groundwork refactor; the AI EXTRACT extension itself will land in a follow-up commit.

done. Please help to review the extension PR :LadybugDB/extensions#60 thx .