Optimize PoW

[poszu]

Part of #416
Optimizes the proof of work:

Single hash benchmark:

BenchmarkPowHash-20    	 2889742	       394.6 ns/op	     152 B/op	       3 allocs/op
↓
BenchmarkPowHash-20    	 3915529	       302.4 ns/op	       0 B/op	       0 allocs/op

Finding PoW benchmark:

BenchmarkFindSubmitPowNonce/difficulty=20-20                   2         625033028 ns/op        243352568 B/op   4802780 allocs/op
↓
BenchmarkFindSubmitPowNonce/difficulty=20-20                   3         348184603 ns/op              416 B/op         6 allocs/op

[codecov]

Codecov Report

Merging #419 (4bba71d) into develop (4bc1a3b) will increase coverage by 0.0%.
The diff coverage is 88.0%.

@@           Coverage Diff           @@
##           develop    #419   +/-   ##
=======================================
  Coverage     75.1%   75.1%           
=======================================
  Files           23      24    +1     
  Lines         1917    1926    +9     
=======================================
+ Hits          1440    1448    +8     
  Misses         359     359           
- Partials       118     119    +1     

Files	Coverage Δ
registration/pow_verifier.go	96.1% <100.0%> (ø)
shared/shared.go	69.7% <ø> (-1.8%)	⬇️
shared/pow.go	87.8% <87.8%> (ø)

... and 1 file with indirect coverage changes

[fasmat]

lgtm

[fasmat]

NIT: h.input does not need to be initialized to append bytes to it

Suggested change

	h := &powHasher{h: ripemd.New256(), input: []byte{}}
	h := &powHasher{h: ripemd.New256()}

[fasmat]

NIT: EqualValues ignores typing; I think this assertion should fail if someone accidentally changes the type of nonce:

Suggested change

	require.EqualValues(t, 143, nonce)
	require.Equal(t, uint64(143), nonce)