Add ladder implementation for inverses modulo the curve order #59

1. 
   sipa commented at 3:37 pm on September 7, 2014: contributor

   This slows down the benchmark (no endomorphism, GMP bignum, 64bit_asm field) by 37%, but removes the need for a secp256k1_num_mod_inverse function.

   EDIT: the signing code now always uses the ladder

2. theuni cross-referenced this on Sep 11, 2014 from issue WIP: Internal bignum by theuni
3. sipa force-pushed on Oct 26, 2014
4. sipa force-pushed on Oct 26, 2014
5. 
   sipa commented at 11:37 am on October 26, 2014: contributor

   @gmaxwell review requested
6. sipa force-pushed on Oct 26, 2014
7. Abstract out inversion modulo order into secp256k1_ge_scalar_inv 2eaa2056c9
8. Implement inverse modulo order using ladder 18bd4a33f7
9. sipa force-pushed on Oct 27, 2014
10. 
    sipa commented at 7:54 am on November 5, 2014: contributor

    Closing, this is pretty outdated with scalar operations outsourced to secp256k1_scalar_t.
11. sipa closed this on Nov 5, 2014

    ---

Contributors
sipa