FFT memory usage estimation

### Description

In cuFFT, the memory requirements for FFT plans can be estimated using the cufftEstimate* functions: https://docs.nvidia.com/cuda/cufft/index.html#cufft-estimated-size-of-work-area

Is there a similar way to estimate memory usage for FFT operations in CuPy, particularly for functions such as: https://docs.cupy.dev/en/stable/reference/generated/cupyx.scipy.fft.ifft2.html#cupyx.scipy.fft.ifft2
?
Are there any plans to add such estimation capabilities to the CuPy API? If not, what would be the recommended approach to implement such estimator functions.

### Additional Information

We may have the resources to add such functionality.

Start with the cuFFT cufftEstimate* documentation and the CuPy cupyx.scipy.fft.ifft2 API referenced in the issue. Determine whether CuPy already exposes a comparable estimation path, then define the estimator scope and recommended API approach. Done means a documented, technically validated proposal or implementation plan for estimating FFT memory usage.

Written by the indexing model from the issue text.