Update primitive-types crate

[NikVolf]

- ancient rand dependency

[NikVolf]

Actually, it seems that it gives like ~5% performance hit during block import
Probably due to paritytech/parity-common#317

And primitive-types/libc feature is explicitly used in sp-core

@ordian ^^^

[NikVolf]

with primitive-types/libc

import block/Wasm       time:   [4.5876 s 4.6357 s 4.7006 s]
import block/Native     time:   [2.5677 s 2.6154 s 2.6846 s]  

without primitive-types/libc:

import block/Wasm       time:   [4.7306 s 4.8279 s 4.9466 s]                               
                        change: [+3.1825% +5.5383% +7.6266%] (p = 0.00 < 0.05)
                        Performance has regressed.

import block/Native     time:   [2.7783 s 2.8249 s 2.8624 s]                                 
                        change: [+2.5561% +6.1032% +9.7396%] (p = 0.00 < 0.05)
                        Performance has regressed.

[ordian]

Actually, it seems that it gives like ~5% performance hit during block import
Probably due to paritytech/parity-common#317

And primitive-types/libc feature is explicitly used in sp-core

@ordian ^^^

I opened rust-lang/rust#69024.

We haven't published a new version of fixed-hash and if you think the libc feature removal should be reverted, we still can do it.

[NikVolf]

We haven't published a new version of fixed-hash and if you think the libc feature removal should be reverted, we still can do it.

Yeah, the fact that it is not published probably unblocks this pr.
Performance issues still should be addressed though - I've benched against just latest parity-common, but we'll figure out later.