Replace direct np.random.* calls with np.random.RandomState instances (#8798)

### Description

Replace direct `np.random.*` global function calls with proper
`np.random.RandomState` instances for reproducibility, as requested in
#6888.

### Changes

| File | Call sites | Fix |
|------|-----------|-----|
| `monai/transforms/utils.py` | 3 × `np.random.random.__self__` | →
`np.random.RandomState()` |
| `monai/transforms/signal/array.py` | 2 × `np.random.choice` | →
`self.R.choice` (classes already Randomizable) |
| `monai/data/synthetic.py` | 2 × `np.random.random.__self__` | →
`np.random.RandomState()` |
| `monai/data/utils.py` | 1 × `np.random.randint` fallback | →
`np.random.RandomState().randint` |
| `monai/utils/ordering.py` | 1 × `np.random.shuffle` | →
`np.random.RandomState().shuffle` |

### Scope

This PR covers 9 functional call sites across 5 files. The remaining
`np.random.*` calls in the codebase are either:
- **Docstring examples** (auto3dseg, visualize) — educational, not
functional
- **`np.random.seed` in `set_determinism()`** — intentionally sets
global state
- **apps/deepedit/ and apps/nuclick/** — would require adding
`Randomizable` inheritance, left for a follow-up

### Checks

- [x] `Signed-off-by` included (DCO)
- [x] No new global random state usage introduced
- [x] Signal transforms use existing `self.R` from `Randomizable` base
class

Ref #6888

---------

Signed-off-by: SexyERIC0723 <haoyuwang144@gmail.com>
Co-authored-by: Eric Kerfoot <17726042+ericspod@users.noreply.github.com>

Author: haoyu-haoyu

monai/transforms/signal/array.py

@@ -273,7 +273,7 @@ def __call__(self, signal: NdarrayOrTensor) -> NdarrayOrTensor:
         data = convert_to_tensor(self.freqs * time_partial)
         sine_partial = self.magnitude * torch.sin(data)
 
-        loc = np.random.choice(range(length))
+        loc = self.R.choice(range(length))
         signal = paste(signal, sine_partial, (loc,))
 
         return signal
@@ -354,7 +354,7 @@ def __call__(self, signal: NdarrayOrTensor) -> NdarrayOrTensor:
         time_partial = np.arange(0, round(self.fracs * length), 1)
         squaredpulse_partial = self.magnitude * squarepulse(self.freqs * time_partial)
 
-        loc = np.random.choice(range(length))
+        loc = self.R.choice(range(length))
         signal = paste(signal, squaredpulse_partial, (loc,))
 
         return signal