fix(harness): add error handling for invoke_harness API call

[jesseturner21]

Summary

• Wraps the invoke_harness_streaming call in a try/except block so boto3 errors (bad credentials, network issues, invalid ARN) produce a clean error message instead of a raw Python traceback in GitHub Actions logs.

Test plan

• Verified Python syntax is valid
• Trigger harness review workflow with an invalid ARN to confirm clean error output

[tejaskash]

LGTM

[github-actions]

Coverage Report

Status	Category	Percentage	Covered / Total
🔵	Lines	43.02%	7794 / 18117
🔵	Statements	42.42%	8247 / 19439
🔵	Functions	40.46%	1349 / 3334
🔵	Branches	40.65%	5108 / 12564

Generated in workflow #2189 for commit 5fd5a0a by the Vitest Coverage Report Action

[agentcore-cli-automation]

Reviewed the change — it's a small, safe addition that wraps the invoke_harness_streaming call in a try/except so boto3 client errors (bad credentials, invalid ARN, etc.) produce a clean error message instead of a raw traceback. Looks good to merge.

One small observation (non-blocking): the try/except only covers the invoke_harness API call itself. Errors that occur while iterating the event stream inside print_stream (e.g. network drops mid-stream, botocore.exceptions.EventStreamError) would still surface as raw tracebacks. Server-side stream errors are already handled via the internalServerException / runtimeClientError event types, so this is only a concern for client-side/transport errors during streaming. Fine to address separately if it ever becomes a problem in practice.