BoxedUint support

[fjarri]

This may supersede #39. It uses *_like constructors (see RustCrypto/crypto-bigint#533) instead of automatic widening, which I think is a safer approach.

The 1024 bit safe prime generation is about 30% slower for BoxedUint compared to Uint. Not too bad for a start, but clearly there is room for improvement.

@xuganyu96 sorry, tried to base it on your PR, but I guess I changed the first commit too much, and it's not assigned to you anymore.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (752bdee) 99.20% compared to head (2b44a1f) 99.23%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #40      +/-   ##
==========================================
+ Coverage   99.20%   99.23%   +0.02%     
==========================================
  Files           9        9              
  Lines        1386     1433      +47     
==========================================
+ Hits         1375     1422      +47     
  Misses         11       11              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[fjarri]

@tarcieri I think this concludes the initial support stage for BoxedUint. If you release another pre version of crypto-bigint, I'll publish a pre version of this crate so that you could try it out in rsa.

[tarcieri]

@fjarri RustCrypto/crypto-bigint#535

[fjarri]

Published v0.6.0-pre.0.

Before 0.6.0 proper I may tweak the public API a little with respect to how bits_precision is passed.

Also filed #41 to investigate the BoxedUint performance.