Smit Lunagariya

@AnirudhDagar Once you review it in your free time let me know if I can proceed over this or I need to add something more.

I have created a sample [cusignal_backend](https://github.com/Smit-create/scipy/blob/uarray-signal-check/scipy/signal/backends/cusignal_backend.py) for testing in colab.

Thanks for the review @AnirudhDagar and @czgdp1807 > probably need to adapt the argument replacers accordingly and define a few more than just the `_h_x_replacer` Yes, will do that in...

@rgommers @AnirudhDagar I created [cupy_backend](https://github.com/Smit-create/scipy/blob/uarray-signal-check/scipy/signal/backends/cupy_backend.py) on the other branch and installed it on the [colab notebook](https://colab.research.google.com/drive/1A89k_3Ym1RGSyQmzXlYBkyHlAT_1_5et?usp=sharing). It seems to be working fine. Please have a look at the notebook and...

There seems to be some issue with `all_of_type`: ```py >>> from scipy._lib.uarray import all_of_type >>> import numpy as np >>> @all_of_type(np.ndarray) ... def f(p): ... return p ... >>> sos_f32...

With the present commits, `__ua_convert__` works fine for the added functions. So, next, I'll look into each function's parameters (optional arguments) which might also need to be dispatched.

@rgommers @czgdp1807 @AnirudhDagar This PR now looks good for a first thorough review.

Hi Ralf, I removed the `GNUC` guards and used `npy_cfloat`/ `npy_cdouble`. There were few errors while building like: ``` scipy/signal/bspline_util.c: In function 'C_IIR_order1': scipy/signal/bspline_util.c:50:16: error: invalid operands to binary *...

The results are different from mine on Apple M1 2020. See: ``` % time lpython b.py lpython b.py 36.32s user 0.77s system 99% cpu 37.266 total % time lpython b.py...

@certik The change is reflected in LPython after the sync. Just added a test so that we can merge this.