Normalize p256

[x87-va]

No description provided.

[codecov-commenter]

Codecov Report

Merging #433 (b5f2f3b) into master (df71290) will increase coverage by 0.27%.
The diff coverage is 90.47%.

@@            Coverage Diff             @@
##           master     #433      +/-   ##
==========================================
+ Coverage   64.70%   64.98%   +0.27%     
==========================================
  Files          28       28              
  Lines        3598     3638      +40     
==========================================
+ Hits         2328     2364      +36     
- Misses       1270     1274       +4     

Impacted Files	Coverage Δ
p256/src/ecdsa.rs	82.60% <0.00%> (-7.87%)	⬇️
p256/src/arithmetic/scalar.rs	82.55% <100.00%> (+1.54%)	⬆️
p256/src/arithmetic/util.rs	82.35% <100.00%> (+15.68%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update df71290...b5f2f3b. Read the comment docs.

[tarcieri]

I think it's a bit excessive to loop in num-bigint just for testing this. k256 does a number of other equivalence tests of the field arithmetic with it.

I think ideally we'd even get rid of those and move all of that sort of testing into a proptests crate or something.

For the purposes of testing this, I think it's fine to use crypto_bigint::UInt to compute the various boundary conditions and then check if they're low/high.

[tarcieri]

I believe this is obsoleted by RustCrypto/signatures#393