feat: adding the bartlett function

[Vlor999]

Description

This PR adds the Bartlett (triangular) window function (mlx.core.bartlett), continuing the effort to expand MLX's signal processing capabilities and feature parity with NumPy.

Implementation Details

• C++: Implemented in mlx/ops.cpp.
  • Optimization (No Branching): Instead of using conditional masking (where or less_equal) to build the triangular shape, this implementation leverages the absolute value identity: $w(n) = 1 - \left| \frac{2n}{M-1} - 1 \right|$. This avoids branching in the compute graph, making it significantly more efficient on the GPU.
  • Constant Folding: The scalar factor 2.0 / (M-1) is computed on the CPU prior to graph construction to minimize graph size.
  • Edge Cases: Explicitly handles M < 1 (empty array) and M = 1 (returns [1.0]), matching numpy.bartlett behavior.
• Python Bindings: Exposed via nanobind in python/src/ops.cpp with nb::sig type hinting and LaTeX documentation.

Test Plan

Verified locally against numpy.bartlett for numerical accuracy and edge cases.

Verification script:

import mlx.core as mx
import numpy as np

# 1. Standard Case
M = 50
mx_bartlett = mx.bartlett(M)
np_bartlett = np.bartlett(M)

assert np.allclose(mx_bartlett, np_bartlett, atol=1e-6), "Mismatch in values"

# 2. Edge Case M=1
assert mx.bartlett(1).item() == 1.0
assert np.bartlett(1).item() == 1.0

# 3. Edge Case M=0
assert mx.bartlett(0).size == 0

print("✅ All verifications passed against numpy.bartlett")

Unit Tests:

Added test_bartlett to python/tests/test_ops.py.

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)

[angeloskath]

👍

[angeloskath]

Let's merge after the tests pass and conflicts are resolved.