gh-35294: Avoid a square root computation in EllipticCurve_field.quadratic_twist
### 📚 Description
The current implementation of `quadratic_twist` computes (possibly
several times) unnecessary finite field square roots, but the only
requirement is to select `D` which has no such square roots, which can
be tested efficiently using PARI (by checking that the norm is not a
quadratic residue).
This solves slowness over large finite field extensions:
```
sage: p = 2**255 - 19
....: K.<a> = GF(p**64)
# Sage 9.8
sage: %timeit EllipticCurve(K, [0,
K.random_element()]).quadratic_twist()
17.8 s ± 8.5 s per loop (mean ± std. dev. of 7 runs, 1 loop each)
# This patch
sage: %timeit EllipticCurve(K, [0,
K.random_element()]).quadratic_twist()
9.69 ms ± 388 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)
```
### 📝 Checklist
- [x] I have made sure that the title is self-explanatory and the
description concisely explains the PR.
- [ ] I have linked an issue or discussion.
- [ ] I have created tests covering the changes.
- [ ] I have updated the documentation accordingly.
### ⌛ Dependencies
URL: https://github.com/sagemath/sage/pull/35294
Reported by: Rémy Oudompheng
Reviewer(s): Lorenz Panny
