sage: adapt eclib test expected output, revert cypari2 update

[collares]

Motivation for this change

eclib was updated in #114425, but the test output changed slightly. Upstream is still on eclib 20190909, so I cannot just import an upstream patch.

Should unbreak sage. I've verified that the test outputs are the same up to minor formatting changes.

[collares]

@ofborg build sage

[collares]

@omasanori @timokau @7c6f434c This is now ready for review. Note that I initially tried to upgrade to eclib 20210226, but this new version causes extra test failures so we must be careful to not let other people update it for the time being. I reported this at JohnCremona/eclib#64, and the conclusion is that Sage requires a patch to set precision parameters properly under eclib 20210226.

The eclib 20210226 extra failures are:

sage -t --long --random-seed=0 /nix/store/k9x5qwik8cdr9vm782v5ha116j097qn4-sage-src-9.2/src/sage/schemes/elliptic_curves/ell_rational_field.py
**********************************************************************
File "/nix/store/k9x5qwik8cdr9vm782v5ha116j097qn4-sage-src-9.2/src/sage/schemes/elliptic_curves/ell_rational_field.py", line 1877, in sage.schemes.elliptic_curves.ell_rational_field.EllipticCurve_rational_field.?
Failed example:
    r, s, G = E.simon_two_descent(); r,s
Expected:
    (8, 8)
Got:
    After 10 attempts at enlargement, giving up!
    --points not proved 5-saturated,
    5-saturation failed!
    Failed to saturate MW basis at primes [ 5 ]
    5 After 10 attempts at enlargement, giving up!
    --points not proved 5-saturated,
    5-saturation failed!
    Failed to saturate MW basis at primes [ 5 ]
    5 After 10 attempts at enlargement, giving up!
    --points not proved 5-saturated,
    5-saturation failed!
    Failed to saturate MW basis at primes [ 5 ]
    (8, 8)
**********************************************************************
File "/nix/store/k9x5qwik8cdr9vm782v5ha116j097qn4-sage-src-9.2/src/sage/schemes/elliptic_curves/ell_rational_field.py", line 1883, in sage.schemes.elliptic_curves.ell_rational_field.EllipticCurve_rational_field.?
Failed example:
    E.simon_two_descent()
Expected:
    (2, 3, [(-1/4 : 2377/8 : 1), (323/4 : 1891/8 : 1)])
Got:
    5 (2, 3, [(-1/4 : 2377/8 : 1), (323/4 : 1891/8 : 1)])
**********************************************************************
1 item had failures:
   2 of  88 in sage.schemes.elliptic_curves.ell_rational_field.EllipticCurve_rational_field.?
    [849 tests, 2 failures, 195.74 s]

[dimpase]

@JohnCremona

[dimpase]

not sure what's prevening Sage's upgrade of eclib, apart from labour shortages.

[JohnCremona]

See https://trac.sagemath.org/ticket/31443

[collares]

I have added a separate commit reverting the cypari2 update that just landed on master, because it is causing test failures on aarch64. Upgrading cypari2 properly will be done on staging now that #115200 has been merged.

[collares]

@SuperSandro2000 Thanks for merging! It's much appreciated.