Enhance proposal task error handling in tbtcpg

[lionakhnazarov]

Modified pkg/tbtcpg/tbtcpg.go - Generate method
Before: If any task's Run method returned an error, the entire proposal generation would fail and return that error.
After:
• If a task fails with an error, it logs the error with context (action name, wallet PKH, error details)
• Continues to the next task in the checklist instead of returning
• Only returns a NoopProposal if all tasks fail or complete without result
• Never returns an error from Generate (always succeeds with either a proposal or Noop)

Impact
• Resilience: Proposal generation continues even if individual tasks fail
• Better coordination: Leaders can still generate proposals (Noop or other actions) even when some tasks encounter errors
• Improved debugging: Errors are logged with full context for investigation
• No coordination failures: Proposal generation always succeeds, preventing leader routine failures that cause follower timeouts
This ensures that coordination messages are always sent, preventing the "coordination message not received on time" errors you were seeing.

[lrsaturnino]

Nice work!

Two things for your evaluation:

1. The retry loop in generateProposal() is now dead code. Since Generate() never returns an error anymore, the 2-attempt retry with 1-min delay never actually retries. It always succeeds on the first try. Evaluate if it makes sense to return an error only when all tasks fail. That way, single-task failures still occur (your change), but if everything blows up (a transient RPC outage), the retry loop has a chance to help.

2. The "all tasks completed without result" log at line 122 now also covers "all tasks errored". Might be worth distinguishing the two so operators don't see "completed without result" when in reality everything failed.

[lrsaturnino]

LGTM