Currently on JDC, if SetCustomMiningJob.Success hasn’t been received, the call to:

channel_manager_data.template_id_to_upstream_job_id.get(tid)

returns None, and as a result, we don't share the result with upstream.
This leads to valid shares being wasted.

Proposed solution:

• Cache the shares submitted before SetCustomMiningJob.Success arrives.
• Once the success message is received, forward those cached shares appropriately.

This would ensure no shares are lost during the delay window.