fix: Charge pipe I/O cycles before scheduler-side copies

[Officeyutong]

What problem does this PR solve?

Issue Number: close #xxx

Problem Summary:
Pipe read/write transferred-byte cycles were added to the participating VM-local cycle counters during process_io(). If a writer stayed blocked in WaitForWrite and the root VM terminated before that writer became runnable again, those pending VM-local cycles could be discarded when non-root VMs were purged.

The host-side pipe copy also happened before the scheduler consumed and enforced the transfer cycles for the current iteration, so large pipe transfers could perform memory-copy work before the remaining cycle budget reflected that cost.

What is changed and how it works?

Proposal: xxx

What's Changed:

Charge pipe transfer cycles directly to the scheduler iteration counter in process_io() before performing the memory copy. process_io() now receives the current remaining cycle limit and returns CyclesExceeded if the transfer would exceed that budget.

The scheduler then consumes the updated iteration_cycles in the same iterate_outer() iteration, so pipe I/O cycles cannot be lost when a blocked VM is later purged. The existing exact-cycle spawn test was updated for the corrected accounting timing.

Related changes

• PR to update owner/repo:
• Need to cherry-pick to the release branch

Check List

Tests

• Unit test
• Integration test
• Manual test (add detailed scripts or steps below)
• No code

Side effects

• Performance regression
• Breaking backward compatibility

[mohanson]

This change affects the script's cycles statistics, which may cause network forks.

[XuJiandong]

The cycles are part of consensus

[Officeyutong]

The cycles are part of consensus

So it's better now to make these changes?

[XuJiandong]

It can be changed in next hard fork.