fix: deflake //rs/state_manager:state_manager_integration (#9173)

## Root Cause

All flaky failures across 8 recent CI runs share the same root cause:
the `wait_for_checkpoint` helper in
`rs/state_manager/tests/common/mod.rs` has a 100-second timeout that is
too tight when many tests (~158) run concurrently on CI. Under parallel
execution, the background hash-computation threads get starved for
CPU/IO, causing `wait_for_checkpoint` to time out with:

```
Checkpoint @n didn't complete in 100s
```

This affects many different tests non-deterministically — some runs had
2 failures, others had up to 21 — because any test calling
`wait_for_checkpoint` can be affected depending on system load.

## Fix

Increase the timeout in `wait_for_checkpoint` from 100s to 300s. The
Bazel test target already has `timeout = "long"` (900s), so 300s is well
within the overall test timeout while providing much more headroom for
slow CI environments.

---

This PR was created following the steps in
`.claude/skills/fix-flaky-tests/SKILL.md`.

Author: basvandijk

rs/state_manager/tests/common/mod.rs

@@ -329,7 +329,7 @@ pub fn modify_encoded_stream_helper<F: FnOnce(StreamSlice) -> Stream>(
 pub fn wait_for_checkpoint(state_manager: &impl StateManager, h: Height) -> CryptoHashOfState {
     use std::time::{Duration, Instant};
 
-    let timeout = Duration::from_secs(100);
+    let timeout = Duration::from_secs(300);
     let started = Instant::now();
     while started.elapsed() < timeout {
         match state_manager.get_state_hash_at(h) {