chore: Make scalar i64 type.

[FeliGame]

Proposed changes

Fix 4238

I try to fix this by toggling scalar decl type to i64.
A smoke test case and its running result shown as below:

$ python3 -c "
import sys
# Import directly from the built .so file
import importlib.util
spec = importlib.util.spec_from_file_location('mlx.core', 'build/core.cpython-312-darwin.so')
mx = importlib.util.module_from_spec(spec)
spec.loader.exec_module(mx)
sys.modules['mlx'] = mx
sys.modules['mlx.core'] = mx

print('Loaded from: build/core.cpython-312-darwin.so')
print()

n = mx.iinfo(mx.int32).max
print('int32 max:', n)
print()

# Test 1: Value parameters: only 4 elements requested, but values need int64
try:
    result = mx.arange(n - 1, n + 3)
    print('Test 1 PASSED:', result)
except Exception as e:
    print('Test 1 FAILED:', type(e).__name__, str(e)[:100])

# Test 2: step larger than int32 max
try:
    result = mx.arange(0, 5, 2**31)
    print('Test 2 PASSED:', result)
except Exception as e:
    print('Test 2 FAILED:', type(e).__name__, str(e)[:100])

# Test 3: value like 2**40
try:
    result = mx.arange(2**40, 2**40 + 3)
    print('Test 3 PASSED:', result)
except Exception as e:
    print('Test 3 FAILED:', type(e).__name__, str(e)[:100])

# Test 4: single-arg arange with int64 value
try:
    result = mx.arange(2**40)
    print('Test 4 PASSED (single arg):', result.shape)
except Exception as e:
    print('Test 4 FAILED:', type(e).__name__, str(e)[:100])

# Test 5: linspace with int64 values
try:
    result = mx.linspace(0, 2**40, 5)
    print('Test 5 PASSED:', result)
except Exception as e:
    print('Test 5 FAILED:', type(e).__name__, str(e)[:100])
"
Loaded from: build/core.cpython-312-darwin.so

int32 max: 2147483647

Test 1 PASSED: array([2147483646, 2147483647, -2147483648, -2147483647], dtype=int32)
Test 2 PASSED: array([0], dtype=int32)
Test 3 PASSED: array([2147483647, 2147483647, 2147483647], dtype=int32)
Test 4 FAILED: ValueError [arange] Maximum size exceeded.
Test 5 PASSED: array([0, 2.74878e+11, 5.49756e+11, 8.24634e+11, 1.09951e+12], dtype=float32)

Checklist

Put an x in the boxes that apply.

• I have read the CONTRIBUTING document
• I have run pre-commit run --all-files to format my code / installed pre-commit prior to committing changes
• I have added tests that prove my fix is effective or that my feature works
• I have updated the necessary documentation (if needed)

[zcbenz]

Can you add tests for the issue?

[FeliGame]

Can you add tests for the issue?

Done.