Fix long-pause resume TTL regression

[code-0-stella]

Fix long-pause resume path before instance storage archives

Summary

Prevents a paused stream from remaining resumable after the safe pause TTL window has elapsed.

This addresses the archive-risk where a stream is paused once, the instance TTL is extended, and the pause can then outlast the safe window without a manual restore. In that case, a normal resume() call can hit the host’s “entry archived” error path instead of failing gracefully with contract logic.

What changed

• Added a single maximum safe pause threshold in the stream TTL module.
• Reused that threshold in the resume guard and force-cancel logic so the protocol consistently rejects overlong pauses.
• Added a regression test covering the resume-after-extended-pause case.

Why

A long pause can leave the stream instance storage effectively archived before the caller can recover it. The contract should fail with a clear, intentional PauseThresholdNotMet error instead of allowing the host to surface an opaque archival failure.

Verification

I added a focused regression test for this scenario. The workspace still contains unrelated existing compile errors outside this task, so I left those unchanged as requested.

Closes #393

[drips-wave]

@code-0-stella Great news! 🎉 Based on an automated assessment of this PR, the linked Wave issue(s) no longer count against your application limits.

You can now already apply to more issues while waiting for a review of this PR. Keep up the great work! 🚀

Learn more about application limits