Exceeding phlo_limit deploy stalls shard

[Timekiller7]

What did you expect to see?

A loop.rho deploy that exceeds phlo_limit fails fast without affecting network liveness. Autopropose keeps producing blocks normally on all validators.

loop.rho:

new stdout(`rho:io:stdout`) in {
  new loop in {
    contract loop(@n) = {
      if (n <= 0) {
        stdout!("done")
      } else {
        loop!(n - 1)
      }
    } |
    loop!(1000000000)
  }
}

What did you see instead?

• The shard became stuck: no new blocks were produced and the autopropose script stalled (no proposals emitted)
• Nodes kept looping the same fork-choice tip messages
• After autopropose service restart (step №4) - failing deploys and proposes to validator, on which loop deploy was sent. Fails with Connection error: <_InactiveRpcError....