Is your feature request related to a problem? Please describe.
When using the TensorBoardImageHandler and passing a series of 2D images to the transforms, the plot_2d_or_3d_image plots each 2D image under a separate tag. When I do the same thing with a series of 3D images, they are plotted as multiple cards in the same tag.

Describe the solution you'd like
It would be really nice if both series of 2D images and series of 3D images behaved the same way. Plotting the series under a single tag is great as it reduces the number of tags in my TensorBoard.

I know that with a series of 3D images, the add_animated_gif function is called. Which plots using writer._get_file_writer().add_summary() for each image. However, the 2D images are plotted using writer.add_image() for each image. I tried figuring out why this happens and couldn't figure it out.

Describe alternatives you've considered
At the moment I just attach another dimension to the front of my 2D images so that they are plotted as 3D GIFs.

Additional context
Here's an example of default series of 2D images behavior:

Here's an example of default series of 3D images behavior: