ethcore: update bn version

[andresilva]

This PR updates the bn dependency to use paritytech/bn#9.

Here are the updated results for alt_bn128 builtin benchmarks:

 name                                 control ns/iter  variable ns/iter  diff ns/iter   diff %  speedup
 alt_bn128_add_cdetrio1               270              303                         33   12.22%   x 0.89
 alt_bn128_add_cdetrio10              1,026            589                       -437  -42.59%   x 1.74
 alt_bn128_add_cdetrio11              13,492           5,096                   -8,396  -62.23%   x 2.65
 alt_bn128_add_cdetrio12              13,547           5,198                   -8,349  -61.63%   x 2.61
 alt_bn128_add_cdetrio13              12,084           4,725                   -7,359  -60.90%   x 2.56
 alt_bn128_add_cdetrio14              3,016            1,269                   -1,747  -57.92%   x 2.38
 alt_bn128_add_cdetrio2               276              292                         16    5.80%   x 0.95
 alt_bn128_add_cdetrio3               282              296                         14    4.96%   x 0.95
 alt_bn128_add_cdetrio4               276              291                         15    5.43%   x 0.95
 alt_bn128_add_cdetrio5               269              287                         18    6.69%   x 0.94
 alt_bn128_add_cdetrio6               1,023            621                       -402  -39.30%   x 1.65
 alt_bn128_add_cdetrio7               1,023            639                       -384  -37.54%   x 1.60
 alt_bn128_add_cdetrio8               1,040            593                       -447  -42.98%   x 1.75
 alt_bn128_add_cdetrio9               1,026            589                       -437  -42.59%   x 1.74
 alt_bn128_add_chfast1                12,006           4,476                   -7,530  -62.72%   x 2.68
 alt_bn128_add_chfast2                12,006           4,634                   -7,372  -61.40%   x 2.59
 alt_bn128_mul_cdetrio1               579,366          223,037               -356,329  -61.50%   x 2.60
 alt_bn128_mul_cdetrio11              577,396          220,626               -356,770  -61.79%   x 2.62
 alt_bn128_mul_cdetrio6               578,070          220,062               -358,008  -61.93%   x 2.63
 alt_bn128_mul_chfast1                130,291          47,492                 -82,799  -63.55%   x 2.74
 alt_bn128_mul_chfast2                288,746          109,929               -178,817  -61.93%   x 2.63
 alt_bn128_mul_chfast3                507,314          194,045               -313,269  -61.75%   x 2.61
 alt_bn128_pairing_empty_data         26               26                           0    0.00%   x 1.00
 alt_bn128_pairing_jeff1              16,266,122       7,059,780           -9,206,342  -56.60%   x 2.30
 alt_bn128_pairing_jeff2              16,059,102       7,059,631           -8,999,471  -56.04%   x 2.27
 alt_bn128_pairing_jeff3              16,108,700       7,054,851           -9,053,849  -56.20%   x 2.28
 alt_bn128_pairing_jeff4              24,188,494       10,648,833         -13,539,661  -55.98%   x 2.27
 alt_bn128_pairing_jeff5              24,205,767       10,559,846         -13,645,921  -56.37%   x 2.29
 alt_bn128_pairing_jeff6              16,112,552       7,070,478           -9,042,074  -56.12%   x 2.28
 alt_bn128_pairing_one_point          8,103,014        3,526,155           -4,576,859  -56.48%   x 2.30
 alt_bn128_pairing_ten_point_match_1  81,001,042       35,447,104         -45,553,938  -56.24%   x 2.29
 alt_bn128_pairing_ten_point_match_2  81,309,432       35,577,506         -45,731,926  -56.24%   x 2.29
 alt_bn128_pairing_ten_point_match_3  16,037,892       7,039,923           -8,997,969  -56.10%   x 2.28
 alt_bn128_pairing_two_point_match_2  16,030,578       7,040,279           -8,990,299  -56.08%   x 2.28
 alt_bn128_pairing_two_point_match_3  16,145,136       7,050,670           -9,094,466  -56.33%   x 2.29
 alt_bn128_pairing_two_point_match_4  16,079,453       7,034,531           -9,044,922  -56.25%   x 2.29

[NikVolf]

byteorder need a bump?

[ordian]

Dumb question: why didn't it work w/o the bump?

TIL: Cargo assumes features are additive, so it can build byteorder once, even though some crates use different feature set.

[andresilva]

@ordian byteorder's u128 support required a nightly compiler on that version. Works in stable in the latest version.