Custom `loop_factory` for `asyncio`

Hello,

I have a use-case where we'd like to run the `ipykernel` main-loop using a specific `asyncio.AbstractEventLoop` implementation (not the stdlib's default one). Since in the `main` branch of `ipykernel` the code was migrated to use `anyio`, I believe this boils down to passing a suitable `loop_factory` value in the `backend_options` passed to `anyio.run` (unless Trio mode is enabled, of course).

Would there be any interest in a patch to allow this, essentially adding a `--IPKernelApp.asyncio_loop_factory` config option (a `DottedObjectName`)?

Thanks!

Start by tracing ipykernel's main-loop startup and the anyio.run call, focusing on how backend_options are passed and how Trio mode is selected. The requested change would expose an IPKernelApp.asyncio_loop_factory DottedObjectName and use it for asyncio mode while preserving Trio behavior; no specific files or tests are named.

Written by the indexing model from the issue text.