refactor(workflow): support configurable per-node timeouts and replace hardcoded 1-year activity timeout

[lokewate]

Current Limitation

In workflow/workflow.go, handleTaskNode currently hardcodes a 1-year StartToCloseTimeout:

nodeCtx := workflow.WithActivityOptions(ctx, workflow.ActivityOptions{
    ActivityID:          nodeInfo.ID,
    StartToCloseTimeout: 24 * time.Hour * 365,
})

This presents two major issues:

• Worker Crash Masking: If a worker crashes mid-execution during a non-heartbeating task, Temporal will not detect worker failure or trigger retries for up to 1 year.
• Lack of SLA Flexibility: All task nodes share the same hardcoded timeout regardless of whether they are fast automated tasks or long-running SLA processes.

Suggested Improvement

• Add optional timeout configuration (timeout, start_to_close_timeout, heartbeat_timeout) to Node / TaskTemplate in dsl.go.
• Update handleTaskNode to parse and apply per-node DSL timeouts with a sensible default fallback (e.g., default StartToCloseTimeout and optional HeartbeatTimeout).

Version

No response

Additional Context

No response