Add gr_poly_add_scalar, gr_poly_sub_scalar, etc.

[rburing]

Let me know if there's any way to shorten the code.

• Closes gr_poly_add_si, gr_poly_sub_si, etc. #2193

[rburing]

I'll add gr_poly_addmul_scalar and gr_poly_submul_scalar to this PR tomorrow.

[fredrik-johansson]

Any idea why the CI is failing?

[rburing]

The test

flint/src/gr/test_ring.c

Line 4261 in 4a89949

gr_test_iter(R, state, "sub: ui/si/fmpz/fmpq", gr_test_sub_type_variants, iters, test_flags);

is failing for polynomials over mpn_mod. After trying to force some output I get:

sub
Ring of polynomials over Integers mod 9619630419041620901435312400118315361684135181482916166167977550805210517601877109410312712138085742792467808192 (mpn)
which: 4
alias: 0
x = 9079397581980565535
y = 618951719031417411676930303
y (op) y (1) = 9619630419041620901435312400118315361684135181482916166167977550805210517601877109409693760428133722962771443424
x (op) y (2) = 926336713898529537677559173422236530738054845141216017338574778030216341092355*x + 9619630419041620901435312400118315361684135073643129497565418372137328575484942659846867307721977005235714916288*x^2 + 1152921504606846975*x^3

Flint exception (General error):
    
FAIL

Not sure what's going on.

[fredrik-johansson]

Not sure if this is the only problem, but you need to normalise the output polynomial in the ui, si, fmpz and fmpq methods in the case where len == 0 and c != 0, the reason being that in finite characteristic, c can still convert to zero.

It's unfortunate that the test code doesn't catch this! I think we need some stronger tests for normalisation.

[fredrik-johansson]

See #2223.

[fredrik-johansson]

Even so, gr_poly is actually designed to give correct results even when polynomials are not normalised. I wonder why that is not working as intended here.

[rburing]

@fredrik-johansson Thanks! This should be good to go now (status was used uninitialized in some cases, now fixed).

[fredrik-johansson]

Great, thank you!