gh-328: efficient resampling in ellipticity_ryden04

[Saransh-cpp]

The code now generates only the required values while resampling making it more efficient. The implementation works with the existing test cases, but given that the function accepts ArrayLike args, the following fails -

In [1]: from glass import ellipticity_ryden04

In [2]: mu, sigma, gamma, sigma_gamma = [-1.85, -2.85], [0.89, 0.89, 0.33], [0.056, 0.02, 0.3, 0.2], [0.222]\

In [3]: ellipticity_ryden04(mu, sigma, gamma, sigma_gamma)
----------------------------------------------------------------------
ValueError                           Traceback (most recent call last)
Cell In[3], line 1
----> 1 ellipticity_ryden04(mu, sigma, gamma, sigma_gamma)

File ~/Code/UCL/glass/glass/shapes.py:143, in ellipticity_ryden04(mu, sigma, gamma, sigma_gamma, size, rng)
    141 # default size if not given
    142 if size is None:
--> 143     size = np.broadcast(mu, sigma, gamma, sigma_gamma).shape
    145 # broadcast all inputs to output shape
    146 # this makes it possible to efficiently resample later
    147 mu = np.broadcast_to(mu, size, subok=True)

ValueError: shape mismatch: objects cannot be broadcast to a single shape.  Mismatch is between arg 0 with shape (2,) and arg 1 with shape (3,).

I am not sure if this is a "valid" case or will scientists never pass in such arguments.

Refs: #328
Changed: resampling in ellipticity_ryden04 is more efficient now

[ntessore]

From the other PR:

Maybe I am picking up the wrong cases (cases that scientists won't use). For instance, this would not be broadcastable -

mu, sigma, gamma, sigma_gamma = [-1.85, -2.85], [0.89, 0.89, 0.33], [0.056, 0.02, 0.3, 0.2], [0.222]

Maybe I am misunderstanding, but this wouldn't work even now, would it? In any case, all of the inputs must be broadcastable to a single shape (or size, if given).

[Saransh-cpp]

but this wouldn't work even now, would it?

Right, it won't.

Maybe I was just overthinking all the possible edge cases haha, thanks!

[Saransh-cpp]

Oops, my bad, thanks for the suggestions!

[ntessore]

Great! 🚀

[paddyroddy]

LGTM

[paddyroddy]

Love a good walrus