[Feature Request] Paged Scroll For Model Selection Instead Of Infinite Scroll

### Is there an existing issue for this?

- [X] I have searched the existing issues and checked the recent builds/commits

### What would your feature do ?

Essentially the model selections become page-based instead of infinite scroll. Additional config to choose the max number of models per page rendered. Model hoarders (me) have tons of models and just clicking on the LORA tab completely freezes the browser and scrolling is incredibly slow. This is a feature oriented towards those with large amounts of models.


### Proposed workflow

When either Textual Inversion, Checkpoints, LORAs or Hypernetworks tab are selected, instead of generating and displaying all model cards at once from the respective directory/ies, it splits them into pages with a page number at the top and it only renders the current page number's respective chunk, the rendered cards are respectively offset from the available model list.

### Additional information

_No response_

No files or tests are named. Start by locating the UI entry points for the Textual Inversion, Checkpoints, LORAs, and Hypernetworks tabs, then trace how their model cards are generated; done means each tab renders only the selected page, supports a configurable maximum per page, and avoids freezing with large model collections.

Written by the indexing model from the issue text.