Support for batched `cupyx.scipy.sparse.linalg.spsolve`

### Description

Currently, `cupy.linalg.solve` supports batch processing. Such functionality for `cupyx.scipy.sparse.linalg.spsolve` would be a great asset and would prevent slow, manual loops.

For example:
`N` being the batch dimension
`A x = b`
, with
- A: Sparse matrix with dimension `(N, M, M)` or as list with `N`-Sparse matrices `(M, M)`.

- b: Dense matrix with dimension `(N, M)` or as list with `N`-Dense vectors `(M)`.
- x: Dense matrix with dimension `(N, M)` or as list with `N`-Dense vectors `(M)`.

### Additional Information

_No response_

Start at the `cupyx.scipy.sparse.linalg.spsolve` entry point named in the issue, then compare its behavior with batched `cupy.linalg.solve`. Done means supporting the listed batched or list-of-matrices forms for A and b and returning corresponding dense solutions.

Written by the indexing model from the issue text.