Incorrect split of columns and rows in a plot for multi label confusion matrix.

[DivineOblivion]

🐛 Bug

The specific values of num_labels in MultilabelConfusionMatrix can throw IndexError in the plot function.

The _get_col_row_split function for num_labels=11 returns rows=3, cols=3, which is not enough for 11 axes.

To Reproduce

import torch
from torchmetrics.classification import MultilabelConfusionMatrix

num_labels = 11  # 17 21
example_preds = torch.rand((100, num_labels))
example_targets = torch.randint(0, 2, (100, num_labels))

conf_matrix = MultilabelConfusionMatrix(num_labels)
conf_matrix(example_preds, example_targets)
conf_matrix.plot()

Traceback

Traceback (most recent call last):
  File "D:\test\test.py", line 10, in <module>
    conf_matrix.plot()
  File "D:\test\.venv\lib\site-packages\torchmetrics\classification\confusion_matrix.py", line 466, in plot
    fig, ax = plot_confusion_matrix(val, ax=ax, add_text=add_text, labels=labels)
  File "C:\Users\Anton\AppData\Local\Programs\Python\Python310\lib\contextlib.py", line 79, in inner
    return func(*args, **kwds)
  File "D:\test\.venv\lib\site-packages\torchmetrics\utilities\plot.py", line 248, in plot_confusion_matrix
    ax = axs[i] if rows != 1 and cols != 1 else axs
IndexError: index 9 is out of bounds for axis 0 with size 9

Environment

• TorchMetrics version (and how you installed TM, e.g. conda, pip, build from source): 1.1.1, pip
• Python & PyTorch Version (e.g., 1.0): 3.10.9, 2.0.1
• Any other relevant information such as OS (e.g., Linux): Windows

[github-actions]

Hi! thanks for your contribution!, great first issue!