Shard hub to reduce lock contention

[FZambia]

The benchmark added here shows 3x difference – but in practice, the effect can be larger a smaller depending on a use-case scenario (channel distribution, connect rate, number of cores etc).

1 shard:

go test -run xxx -bench BenchmarkHub_Contention
goos: darwin
goarch: amd64
pkg: github.com/centrifugal/centrifuge
cpu: Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz
BenchmarkHub_Contention-8   	    3283	    411077 ns/op

64 shards:

go test -run xxx -bench BenchmarkHub_Contention
goos: darwin
goarch: amd64
pkg: github.com/centrifugal/centrifuge
cpu: Intel(R) Core(TM) i7-7700HQ CPU @ 2.80GHz
BenchmarkHub_Contention-8   	    9630	    147781 ns/op

[codecov]

Codecov Report

Merging #184 (d1d022b) into master (aa6f7c3) will increase coverage by 0.09%.
The diff coverage is 96.87%.

@@            Coverage Diff             @@
##           master     #184      +/-   ##
==========================================
+ Coverage   83.16%   83.26%   +0.09%     
==========================================
  Files          33       33              
  Lines        4361     4393      +32     
==========================================
+ Hits         3627     3658      +31     
- Misses        467      471       +4     
+ Partials      267      264       -3     

Impacted Files	Coverage Δ
hub.go	85.15% <96.87%> (+4.28%)	⬆️
broker_redis.go	73.12% <0.00%> (-0.86%)	⬇️
metrics.go	86.53% <0.00%> (-0.68%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update aa6f7c3...d1d022b. Read the comment docs.