Add explicit index and bytes support

[aaishwarymishra]

• ☑️ I understand it is strictly prohibited to use AI to write PR description
  Hi, this PR adds __index__ support to MLX arrays, allowing integer-valued scalar arrays to be used directly as Python indices. I found MLX did not had __index__ in data-apis/array-api-compat#451, adding __index__ caused regressions in boolean masking and bytes() conversion. Fixed boolean masking indexing behavior and added an explicit __bytes__ implementation.

• AI usage disclosure:
  AI was used for bug Identification and diagnosis caused by introducing __index__.

[ev-br]

From the Array API perspective, having the __index__ dunder to match the existing __int__ dunder would be a great QoL improvement, as it enables the idiomatic operator.index usage, per PEP 357:

In [1]: import mlx.core as mx

In [2]: int(mx.asarray(2))
Out[2]: 2

In [5]: import operator

In [6]: operator.index(mx.asarray(2))
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[6], line 1
----> 1 operator.index(mx.asarray(2))

TypeError: 'mlx.core.array' object cannot be interpreted as an integer