Add batch auto-balancer relaunch and supervisor admin tx on v0

[liobrasil]

Summary

• add a single admin transaction to batch relaunch multiple auto-balancers by id
• make the relaunch logic stuck-aware so it only cancels live scheduled entries when needed
• ensure the global supervisor is present and schedule a fresh recurring supervisor run in the same transaction

Why

The current admin flow requires multiple transactions:

• relaunch each auto-balancer one by one
• reason manually about whether cancellation is safe based on stuck state
• separately schedule or reset the supervisor

This PR adds one v0 admin entrypoint that can do the full recovery pass atomically.

Behavior

For each requested auto-balancer id:

• dedupe ids in the input array
• borrow the auto-balancer from storage
• if the vault is not stuck, cancel only Scheduled transactions so the new config takes effect immediately
• if the vault is stuck, leave inactive scheduled records alone and seed the next rebalance directly
• apply a new recurring config and immediately call scheduleNextRebalance(nil)

For the supervisor:

• detect an invalid stored supervisor capability and reset that state before reconfiguring
• call ensureSupervisorConfigured() so a missing supervisor is recreated
• cancel any currently scheduled supervisor run
• schedule a fresh recurring supervisor execution with the provided interval / effort / scan settings

Validation

• flow cadence lint cadence/transactions/flow-yield-vaults/admin/batch_relaunch_auto_balancers_and_schedule_supervisor.cdc --warnings-as-errors
  • passed
• flow cadence lint cadence/transactions/flow-yield-vaults/admin/relaunch_auto_balancer_by_id.cdc
  • passed in the same bootstrapped origin/v0 worktree

Notes

This is scoped to the v0 architecture and uses:

• FlowYieldVaultsAutoBalancersV1
• FlowYieldVaultsSchedulerV1
• AutoBalancers.AutoBalancerRecurringConfig