feat: add ability to stop in-progress ingestion jobs

[nborges-aws]

Description

Users have no way to cancel a running ingestion job through the CLI. Currently, if a user starts ingestion on a large data source and realizes they made a config mistake, they're forced to wait for the full job to complete before re-ingesting with the corrected config. This wastes compute (driving up cost) and delays their workflow.

The Bedrock API supports StopIngestionJob — we should expose this through the CLI.

Acceptance Criteria

• User can stop an in-progress ingestion job for a named knowledge base
• Command reports which job(s) were stopped and their final status
• Command errors clearly if no ingestion job is currently running for the specified KB
• Works for both single-DS and multi-DS knowledge bases (stops all active jobs)

Additional Context

Decision needed on command surface:

• Option 1: Flag on run — agentcore run ingest --name <kb> --stop
• Option 2: Top-level stop — agentcore stop ingest --name <kb>
• Option 3: Job engine — Register ingestion as a JobType with Stoppable trait