Failing test scenarios for ttnn.max

[bbradelTT]

Internal commits for #16163 have attempted fix that didn't work because of downstream issues that need to be addressed. Should be a basis for new PRs.

Need to look at 3 parts:

1. add more shapes to tests besides what's listed below

2. look at failures (with release build) for

FAILED tests/sweep_framework/sweeps/reduction/traces/max_traces.py::test_forge[layout1-dtype0-params2] - RuntimeError: TT_FATAL @ ../ttnn/cpp/ttnn/tensor/layout/page_config.cpp:109: (heightAlignment % tile_.get_height()) == 0

3. Fix the following.
   For

import pytest
import torch
import ttnn
from tests.ttnn.utils_for_testing import assert_with_pcc
from models.utility_functions import torch_random

@pytest.mark.parametrize("input_shape_and_dim", 
                         [
                             ((2, 32, 64, 64),-4),
                             ((2, 22, 37, 41),-4),
                             ((2, 32, 64, 64),-3),
                             ((2, 22, 37, 41),-3),
                             ((2, 32, 64),-3),
                             ((2, 22, 37),-3)
                         ])
def test_max_unsupported_dim(device, input_shape_and_dim):
    
    input_shape, max_dim = input_shape_and_dim
    
    torch_input_tensor = torch_random(input_shape, -100, 100, dtype=torch.bfloat16)
    torch_output_tensor, _ = torch.max(torch_input_tensor, dim=max_dim, keepdim=True)

    input_tensor = ttnn.from_torch(torch_input_tensor, layout=ttnn.TILE_LAYOUT, device=device)

    output_tensor = ttnn.max(input_tensor, dim=max_dim)
    output_tensor = ttnn.to_layout(output_tensor, ttnn.TILE_LAYOUT)
    output_tensor = ttnn.from_device(output_tensor)

    output_tensor = ttnn.to_torch(output_tensor)

    assert_with_pcc(torch_output_tensor, output_tensor)

All tests except the first fail even after indices are adjusted. Output is something like

PASSED test_mean.py::test_max_unsupported_dim[input_shape_and_dim2]
FAILED test_mean.py::test_max_unsupported_dim[input_shape_and_dim0] - AssertionError: 0.9424668632904696
FAILED test_mean.py::test_max_unsupported_dim[input_shape_and_dim1] - AssertionError: list(expected_pytorch_result.shape)=[1, 22, 37, 41] vs list(actual_pytorch_result.shape)=[1, 22, 64, 64]
FAILED test_mean.py::test_max_unsupported_dim[input_shape_and_dim3] - AssertionError: list(expected_pytorch_result.shape)=[2, 1, 37, 41] vs list(actual_pytorch_result.shape)=[2, 1, 64, 64]
FAILED test_mean.py::test_max_unsupported_dim[input_shape_and_dim4] - RuntimeError: TT_FATAL @ ../ttnn/cpp/ttnn/operations/data_movement/pad/pad.cpp:41: rank == 4
FAILED test_mean.py::test_max_unsupported_dim[input_shape_and_dim5] - RuntimeError: TT_FATAL @ ../ttnn/cpp/ttnn/operations/data_movement/pad/pad.cpp:41: rank == 4

These failing scenarios need to be fixed.

[bbradelTT]

Should be fixed by #16989